Upload 18 files

script/CogVLM2-0-shot.py

@@ -0,0 +1,99 @@
+import os
+import torch
+from PIL import Image
+from modelscope import AutoModelForCausalLM, AutoTokenizer
+
+MODEL_PATH = "./cogvlm2-llama3-chat-19B-int4"
+DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'
+TORCH_TYPE = torch.bfloat16 if torch.cuda.is_available() and torch.cuda.get_device_capability()[0] >= 8 else torch.float16
+
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH, trust_remote_code=True)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_PATH,
+    torch_dtype=TORCH_TYPE,
+    trust_remote_code=True,
+    low_cpu_mem_usage=True,
+).eval()
+
+
+question = (
+''' "1. Identify the type of fruit or crop shown in the image.  "
+    "2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  "
+    "3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, picking it, discard it)  "
+    "4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely). "
+    "Please respond in the following format:\n"
+    "Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  "
+    "Recommendation: [keep for further growth / try to recover it / picking it / discard it]  Consumer Score: [1-100]'''
+)
+
+
+root_folder = "../data"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.jpg', '.jpeg', '.png', '.bmp', '.webp'))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+ 
+
+        with open(output_file, "w", encoding="utf-8") as out_file:
+            for filename in image_files:
+                image_path = os.path.join(subfolder_path, filename)
+                try:
+                    print(f"🖼️ Processing {filename}...")
+
+                    image = Image.open(image_path).convert('RGB')
+                    history = []
+
+                    input_by_model = model.build_conversation_input_ids(
+                        tokenizer,
+                        query=question,
+                        history=history,
+                        images=[image],
+                        template_version='chat'
+                    )
+
+                    inputs = {
+                        'input_ids': input_by_model['input_ids'].unsqueeze(0).to(DEVICE),
+                        'token_type_ids': input_by_model['token_type_ids'].unsqueeze(0).to(DEVICE),
+                        'attention_mask': input_by_model['attention_mask'].unsqueeze(0).to(DEVICE),
+                        'images': [[input_by_model['images'][0].to(DEVICE).to(TORCH_TYPE)]],
+                    }
+
+                    gen_kwargs = {
+                        "max_new_tokens": 2048,
+                        "pad_token_id": 128002,
+                    }
+
+                    with torch.no_grad():
+                        outputs = model.generate(**inputs, **gen_kwargs)
+                        outputs = outputs[:, inputs['input_ids'].shape[1]:]
+                        response = tokenizer.decode(outputs[0])
+                        response = response.split("<|end_of_text|>")[0].strip()
+
+                    print(f"✅ Response: {response}\n")
+                    out_file.write(f"{'='*25} IMAGE START {'='*25}\n")
+                    out_file.write(f"🖼️ Image Name: {filename}\n")
+                    out_file.write(f"📝 Answer:\n{response}\n")
+                    out_file.write(f"{'='*25} IMAGE END {'='*25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{filename}: {e}")
+                    out_file.write(f"{'='*25} IMAGE START {'='*25}\n")
+                    out_file.write(f"🖼️ Image Name: {filename}\n")
+                    out_file.write(f"❌ ERROR: {e}\n")
+                    out_file.write(f"{'='*25} IMAGE END {'='*25}\n\n")

script/DeepSeekVL-chat-0-shot.py

@@ -0,0 +1,111 @@
+import os
+import torch
+from tqdm import tqdm
+from transformers import AutoModelForCausalLM
+from deepseek_vl.models import VLChatProcessor, MultiModalityCausalLM
+from deepseek_vl.utils.io import load_pil_images
+from PIL import Image
+
+
+model_path = "deepseek-ai/deepseek-vl-7b-chat"
+vl_chat_processor: VLChatProcessor = VLChatProcessor.from_pretrained(model_path)
+tokenizer = vl_chat_processor.tokenizer
+
+vl_gpt: MultiModalityCausalLM = AutoModelForCausalLM.from_pretrained(
+    model_path,
+    cache_dir=".",
+    trust_remote_code=True
+)
+vl_gpt = vl_gpt.to(torch.bfloat16).cuda().eval()
+
+
+question = (
+    "1. Identify the type of fruit or crop shown in the image.  "
+    "2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  "
+    "3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  "
+    "4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely). "
+    "Please respond in the following format:\n"
+    "Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  "
+    "Recommendation: [keep for further growth / try to recover it /picking it/ discard it]  Consumer Score: [1-100]"
+)
+
+
+root_folder = "../data/"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.jpg', '.jpeg', '.png', '.bmp'))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+   
+
+        with open(output_file, "w", encoding="utf-8") as fout:
+            for image_file in tqdm(image_files, desc=f"{fruit}/{subfolder}"):
+                image_path = os.path.join(subfolder_path, image_file)
+
+                try:
+             
+                    conversation = [
+                        {
+                            "role": "User",
+                            "content": "<image_placeholder>" + question,
+                            "images": [image_path],
+                        },
+                        {"role": "Assistant", "content": ""}
+                    ]
+
+               
+                    pil_images = load_pil_images(conversation)
+
+             
+                    prepare_inputs = vl_chat_processor(
+                        conversations=conversation,
+                        images=pil_images,
+                        force_batchify=True
+                    ).to(vl_gpt.device)
+
+            
+                    inputs_embeds = vl_gpt.prepare_inputs_embeds(**prepare_inputs)
+
+     
+                    outputs = vl_gpt.language_model.generate(
+                        inputs_embeds=inputs_embeds,
+                        attention_mask=prepare_inputs.attention_mask,
+                        pad_token_id=tokenizer.eos_token_id,
+                        bos_token_id=tokenizer.bos_token_id,
+                        eos_token_id=tokenizer.eos_token_id,
+                        max_new_tokens=512,
+                        do_sample=False,
+                        use_cache=True
+                    )
+
+          
+                    answer = tokenizer.decode(outputs[0].cpu().tolist(), skip_special_tokens=True).strip()
+
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"📝 Answer:\n{answer}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+                    print(f"✅ {image_file} => {answer.splitlines()[0]}")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{image_file}: {e}")
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"❌ ERROR: {e}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+

script/DeepSeekVL-chat-1-shot.py

@@ -0,0 +1,135 @@
+# -*- coding: utf-8 -*-
+
+
+import os, gc, torch
+from tqdm import tqdm
+from transformers import AutoModelForCausalLM
+from deepseek_vl.models import VLChatProcessor, MultiModalityCausalLM
+from deepseek_vl.utils.io import load_pil_images
+
+
+chinese_to_english = {
+    'strawberry', 'tomato', 'guava', 'dragon fruit',
+    'orange',   'pear',     'lychee',   'mango',
+    'kiwi',    'papaya', 'apple',    'grape',
+    'pomegranate','peach', 'banana',   'pomelo'
+}
+
+stage_to_english = {
+    'unripe', 'mature', 'pest-damaged', 'rotten'
+}
+
+recommendation_map = {
+    'unripe': 'keep for further growth',
+    'mature': 'picking it',
+    'pest-damaged': 'try to recover it',
+    'rotten': 'discard it'
+}
+
+score_map = {
+    'unripe': 30, 'mature': 85, 'pest-damaged': 20, 'rotten': 5
+}
+
+
+model_path  = "deepseek-ai/deepseek-vl-7b-chat"   
+root_folder = "../data/水果"                       
+output_root = "result_1shot"
+os.makedirs(output_root, exist_ok=True)
+
+print("🚀 Loading DeepSeek‑VL‑7B ...")
+vl_chat_processor: VLChatProcessor = VLChatProcessor.from_pretrained(model_path)
+tokenizer = vl_chat_processor.tokenizer
+vl_gpt: MultiModalityCausalLM = AutoModelForCausalLM.from_pretrained(
+    model_path, cache_dir=".", trust_remote_code=True
+).to(torch.bfloat16).cuda().eval()
+
+
+
+question = (
+    "1. Identify the type of fruit or crop shown in the image.  \n"
+    "2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  \n"
+    "3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  \n"
+    "4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).\n"
+    "Please respond in the following format:\n"
+    "Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  "
+    "Recommendation: [keep for further growth / pick it /try to recover it / discard it]  Consumer Score: [1-100]"
+)
+
+
+def build_stage_example(fruit_dir: str, fruit_cn: str, stage_cn: str):
+    stage_path = os.path.join(fruit_dir, stage_cn)
+    imgs = sorted([f for f in os.listdir(stage_path)
+                   if f.lower().endswith((".png", ".jpg", ".jpeg", ".bmp"))])
+    if not imgs:
+        return []
+
+    img_path = os.path.join(stage_path, imgs[0])  
+
+    fruit_en = chinese_to_english.get(fruit_cn, fruit_cn)
+    stage_en = stage_to_english.get(stage_cn, "mature")
+    assistant_reply = (
+        f"Type: {fruit_en}  Growth Stage: {stage_en}  "
+        f"Recommendation: {recommendation_map[stage_en]}  "
+        f"Consumer Score: {score_map[stage_en]}"
+    )
+
+
+    return [
+        {"role": "User", "content": "<image_placeholder>\n" + question, "images": [img_path]},
+        {"role": "Assistant", "content": assistant_reply},
+    ]
+
+
+for fruit_cn in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit_cn)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for stage_cn in os.listdir(fruit_path):
+        stage_path = os.path.join(fruit_path, stage_cn)
+        if not os.path.isdir(stage_path):
+            continue
+
+        img_files = sorted([f for f in os.listdir(stage_path)
+                            if f.lower().endswith((".png", ".jpg", ".jpeg", ".bmp"))])
+      
+
+        example_msgs = build_stage_example(fruit_path, fruit_cn, stage_cn)
+        if not example_msgs:
+            continue
+
+       
+        output_file = os.path.join(output_root, f"{fruit_cn}_{stage_cn}.txt")
+        with open(output_file, "w", encoding="utf-8") as fout:
+            for img_name in tqdm(img_files, desc=f"{fruit_cn}/{stage_cn}"):
+                img_path = os.path.join(stage_path, img_name)
+                if img_path in example_msgs[0]["images"]:
+                    continue  
+
+                conversation = example_msgs + [
+                    {"role": "User", "content": "<image_placeholder>\n" + question, "images": [img_path]},
+                    {"role": "Assistant", "content": ""},
+                ]
+                try:
+                    with torch.no_grad():
+                        pil_imgs = load_pil_images(conversation)
+                        prep_in  = vl_chat_processor(conversations=conversation, images=pil_imgs, force_batchify=True).to(vl_gpt.device)
+                        embeds   = vl_gpt.prepare_inputs_embeds(**prep_in)
+                        out_ids  = vl_gpt.language_model.generate(inputs_embeds=embeds, attention_mask=prep_in.attention_mask,
+                                                                  pad_token_id=tokenizer.eos_token_id,
+                                                                  bos_token_id=tokenizer.bos_token_id,
+                                                                  eos_token_id=tokenizer.eos_token_id,
+                                                                  max_new_tokens=256, do_sample=False, use_cache=True)
+                        answer   = tokenizer.decode(out_ids[0], skip_special_tokens=True).strip()
+
+                    print(f"✅ {img_name} → {answer.splitlines()[0]}")
+                    fout.write("="*25 + " IMAGE START " + "="*25 + "\n")
+                    fout.write(f"🖼️ Image Name: {img_name}\n")
+                    fout.write(f"📝 Answer:\n{answer}\n")
+                    fout.write("="*25 + " IMAGE END " + "="*25 + "\n\n")
+                except Exception as e:
+                    print(f"[ERROR] {fruit_cn}/{stage_cn}/{img_name}: {e}")
+
+                finally:
+                    torch.cuda.empty_cache()
+                    gc.collect()

script/DeepSeekVL2-0-shot.py.py

@@ -0,0 +1,107 @@
+import torch,os
+from tqdm import tqdm
+from transformers import AutoModelForCausalLM
+
+from deepseek_vl2.models import DeepseekVLV2Processor, DeepseekVLV2ForCausalLM
+from deepseek_vl2.utils.io import load_pil_images
+
+#https://huggingface.co/deepseek-ai/deepseek-vl-7b-chat
+
+# specify the path to the model
+model_path = "deepseek-ai/deepseek-vl2-tiny"
+# model_path = "deepseek-ai/deepseek-vl2-small"
+vl_chat_processor: DeepseekVLV2Processor = DeepseekVLV2Processor.from_pretrained(model_path,cache_dir=".")
+tokenizer = vl_chat_processor.tokenizer
+
+vl_gpt: DeepseekVLV2ForCausalLM = AutoModelForCausalLM.from_pretrained(model_path,cache_dir=".", trust_remote_code=True)
+vl_gpt = vl_gpt.to(torch.bfloat16).cuda().eval()
+
+image_folder = "../images"
+result_file = "./results.txt"
+
+
+user_prompt = (
+    "This is the image: <image>\n"
+    "1. Identify the type of fruit or crop shown in the image.  "
+    "2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  "
+    "3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  "
+    "4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely). "
+    "Please respond in the following format:\n"
+    "Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  "
+    "Recommendation: [keep for further growth / try to recover it /picking it/ discard it]  Consumer Score: [1-100]"
+
+)
+
+
+root_folder = "../data"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for status in os.listdir(fruit_path):
+        status_path = os.path.join(fruit_path, status)
+        if not os.path.isdir(status_path):
+            continue
+
+        image_files = [f for f in os.listdir(status_path) if f.lower().endswith((".jpg", ".jpeg", ".png"))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{status}.txt")
+       
+        results = []
+
+        for img_name in tqdm(image_files, desc=f"{fruit}/{status}"):
+            img_path = os.path.join(status_path, img_name)
+
+            conversation = [
+                {"role": "<|User|>", "content": user_prompt, "images": [img_path]},
+                {"role": "<|Assistant|>", "content": ""}
+            ]
+
+            try:
+                pil_images = load_pil_images(conversation)
+                prepare_inputs = vl_chat_processor(
+                    conversations=conversation,
+                    images=pil_images,
+                    force_batchify=True,
+                    system_prompt=""
+                ).to(vl_gpt.device)
+
+                inputs_embeds = vl_gpt.prepare_inputs_embeds(**prepare_inputs)
+
+                outputs = vl_gpt.language.generate(
+                    inputs_embeds=inputs_embeds,
+                    attention_mask=prepare_inputs.attention_mask,
+                    pad_token_id=tokenizer.eos_token_id,
+                    bos_token_id=tokenizer.bos_token_id,
+                    eos_token_id=tokenizer.eos_token_id,
+                    max_new_tokens=512,
+                    do_sample=False,
+                    use_cache=True
+                )
+
+                decoded = tokenizer.decode(outputs[0], skip_special_tokens=False)
+                assistant_reply = decoded.split("<|Assistant|>")[-1].strip()
+
+                result_entry = f"{'='*25} IMAGE START {'='*25}\n"
+                result_entry += f"🖼️ Image Name: {img_name}\n"
+                result_entry += f"📝 Answer:\n{assistant_reply}\n"
+                result_entry += f"{'='*25} IMAGE END {'='*25}\n\n"
+                results.append(result_entry)
+
+            except Exception as e:
+                print(f"[ERROR] {fruit}/{status}/{img_name}: {e}")
+                results.append(f"{'='*25} IMAGE START {'='*25}\n")
+                results.append(f"🖼️ Image Name: {img_name}\n")
+                results.append(f"❌ ERROR: {e}\n")
+                results.append(f"{'='*25} IMAGE END {'='*25}\n\n")
+
+       
+        with open(output_file, "w", encoding="utf-8") as fout:
+            fout.writelines(results)

script/DeepSeekVL2-1-shot.py

@@ -0,0 +1,151 @@
+# -*- coding: utf-8 -*-
+"""
+1-shot 推理脚本（每类水果、每张图片使用其真实阶段的示例图）
+--------------------------------
+- 每张图对应一个同阶段的示例图（取文件名排序后的第一张）+ 示例回答
+- 检查 token 长度是否超出限制（默认 4096）
+- 支持 CPU 或 GPU 部署
+- 推理结果保存在 result_1shot/ 目录
+- 每轮推理后自动清理 CUDA 显存与垃圾回收
+"""
+
+import os
+import gc
+import torch
+from tqdm import tqdm
+from transformers import AutoModelForCausalLM
+from DeepSeek_VL2.deepseek_vl2.models import DeepseekVLV2Processor, DeepseekVLV2ForCausalLM
+from DeepSeek_VL2.deepseek_vl2.utils.io import load_pil_images
+
+chinese_to_english = {
+    'strawberry', 'tomato', 'guava', 'dragon fruit',
+    'orange',   'pear',     'lychee',   'mango',
+    'kiwi',    'papaya', 'apple',    'grape',
+    'pomegranate','peach', 'banana',   'pomelo'
+}
+
+stage_to_english = {
+    'unripe', 'mature', 'pest-damaged', 'rotten'
+}
+
+recommendation_map = {
+    'unripe': 'keep for further growth',
+    'mature': 'picking it',
+    'pest-damaged': 'try to recover it',
+    'rotten': 'discard it'
+}
+
+score_map = {
+    'unripe': 30, 'mature': 85, 'pest-damaged': 20, 'rotten': 5
+}
+
+max_token_limit = 4096
+
+model_path = "deepseek-ai/deepseek-vl2-tiny"
+use_gpu = torch.cuda.is_available()
+
+print(f"🚀 Loading model ({'GPU' if use_gpu else 'CPU'}) ...")
+vl_proc: DeepseekVLV2Processor = DeepseekVLV2Processor.from_pretrained(model_path, cache_dir=".")
+tokenizer = vl_proc.tokenizer
+vl_gpt: DeepseekVLV2ForCausalLM = AutoModelForCausalLM.from_pretrained(
+    model_path, cache_dir=".", trust_remote_code=True
+)
+if use_gpu:
+    vl_gpt = vl_gpt.to(torch.bfloat16).cuda().eval()
+else:
+    vl_gpt = vl_gpt.eval()
+print("✅ Model ready.\n")
+
+user_prompt = (
+    "This is the image: <image>\n"
+    "1. Identify the type of fruit or crop shown in the image.  "
+    "2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  "
+    "3. Recommend the farmer’s next action. (Options: keep for further growth, picking it,try to recover it,  discard it)  "
+    "4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely). "
+    "Please respond in the following format:\n"
+    "Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  "
+    "Recommendation: [keep for further growth / picking it /try to recover it /  discard it]  Consumer Score: [1-100]"
+)
+
+root_folder = "../data"
+output_root = "result_1shot"
+os.makedirs(output_root, exist_ok=True)
+
+for fruit_cn in os.listdir(root_folder):
+    fruit_dir = os.path.join(root_folder, fruit_cn)
+    if not os.path.isdir(fruit_dir):
+        continue
+
+    fruit_en = chinese_to_english.get(fruit_cn, fruit_cn)
+
+    for stage_cn in os.listdir(fruit_dir):
+        stage_dir = os.path.join(fruit_dir, stage_cn)
+        if not os.path.isdir(stage_dir):
+            continue
+
+        image_files = sorted([f for f in os.listdir(stage_dir) if f.lower().endswith(('.jpg', '.jpeg', '.png'))])
+        
+        sample_file = image_files[0]
+        sample_path = os.path.join(stage_dir, sample_file)
+        stage_en = stage_to_english.get(stage_cn, 'mature')
+        recomm = recommendation_map.get(stage_en, 'picking it')
+        score = score_map.get(stage_en, 85)
+        sample_answer = f"Type: {fruit_en}  Growth Stage: {stage_en}  Recommendation: {recomm}  Consumer Score: {score}"
+        
+        output_file = os.path.join(output_root, f"{fruit_cn}_{stage_cn}.txt")
+        
+        for image_file in tqdm(image_files, desc=f"{fruit_cn}/{stage_cn}"):
+            image_path = os.path.join(stage_dir, image_file)
+            if image_file == sample_file:
+                continue  
+
+            try:
+                conversation = [
+                    {"role": "<|User|>", "content": "This is the image: <image>\n1. Identify the fruit.\n2. Current growth stage.\n3. Recommended action.\n4. Consumer willingness score.", "images": [sample_path]},
+                    {"role": "<|Assistant|>", "content": sample_answer},
+                    {"role": "<|User|>", "content": user_prompt, "images": [image_path]},
+                    {"role": "<|Assistant|>", "content": ""}
+                ]
+
+                pil_images = load_pil_images(conversation)
+                inputs = vl_proc(conversations=conversation, images=pil_images, force_batchify=True, system_prompt="")
+                if use_gpu:
+                    inputs = inputs.to(vl_gpt.device)
+
+                num_tokens = inputs.input_ids.shape[1]
+                
+
+                inp_emb = vl_gpt.prepare_inputs_embeds(**inputs)
+                gen_out = vl_gpt.language.generate(
+                    inputs_embeds=inp_emb,
+                    attention_mask=inputs.attention_mask,
+                    pad_token_id=tokenizer.eos_token_id,
+                    bos_token_id=tokenizer.bos_token_id,
+                    eos_token_id=tokenizer.eos_token_id,
+                    max_new_tokens=512,
+                    do_sample=False,
+                    use_cache=True
+                )
+
+                reply = tokenizer.decode(gen_out[0]).split("<|Assistant|>")[-1].strip()
+
+                with open(output_file, "a", encoding="utf-8") as fout:
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"📝 Answer:\n{reply}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+                print(f"✅ {image_file} => {reply.splitlines()[0]}")
+
+            except Exception as e:
+                print(f"[ERROR] {fruit_cn}/{stage_cn}/{image_file}: {e}")
+                with open(output_file, "a", encoding="utf-8") as fout:
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"❌ ERROR: {e}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+            finally:
+                if use_gpu:
+                    torch.cuda.empty_cache()
+                    gc.collect()

script/InterVL2.5-4B-0-shot.py

@@ -0,0 +1,135 @@
+import os
+import torch
+from tqdm import tqdm
+from PIL import Image
+from transformers import AutoModel, AutoTokenizer
+import torchvision.transforms as T
+from torchvision.transforms.functional import InterpolationMode
+
+
+IMAGENET_MEAN = (0.485, 0.456, 0.406)
+IMAGENET_STD = (0.229, 0.224, 0.225)
+
+
+def build_transform(input_size):
+    return T.Compose([
+        T.Lambda(lambda img: img.convert('RGB') if img.mode != 'RGB' else img),
+        T.Resize((input_size, input_size), interpolation=InterpolationMode.BICUBIC),
+        T.ToTensor(),
+        T.Normalize(mean=IMAGENET_MEAN, std=IMAGENET_STD)
+    ])
+
+def find_closest_aspect_ratio(aspect_ratio, target_ratios, width, height, image_size):
+    best_ratio_diff = float('inf')
+    best_ratio = (1, 1)
+    area = width * height
+    for ratio in target_ratios:
+        target_aspect_ratio = ratio[0] / ratio[1]
+        ratio_diff = abs(aspect_ratio - target_aspect_ratio)
+        if ratio_diff < best_ratio_diff:
+            best_ratio_diff = ratio_diff
+            best_ratio = ratio
+        elif ratio_diff == best_ratio_diff:
+            if area > 0.5 * image_size * image_size * ratio[0] * ratio[1]:
+                best_ratio = ratio
+    return best_ratio
+
+def dynamic_preprocess(image, min_num=1, max_num=12, image_size=448, use_thumbnail=False):
+    orig_width, orig_height = image.size
+    aspect_ratio = orig_width / orig_height
+    target_ratios = sorted(
+        [(i, j) for n in range(min_num, max_num + 1) for i in range(1, n + 1)
+         for j in range(1, n + 1) if i * j <= max_num and i * j >= min_num],
+        key=lambda x: x[0] * x[1]
+    )
+    target_aspect_ratio = find_closest_aspect_ratio(aspect_ratio, target_ratios, orig_width, orig_height, image_size)
+    target_width = image_size * target_aspect_ratio[0]
+    target_height = image_size * target_aspect_ratio[1]
+    blocks = target_aspect_ratio[0] * target_aspect_ratio[1]
+    resized_img = image.resize((target_width, target_height))
+    processed_images = []
+    for i in range(blocks):
+        box = (
+            (i % (target_width // image_size)) * image_size,
+            (i // (target_width // image_size)) * image_size,
+            ((i % (target_width // image_size)) + 1) * image_size,
+            ((i // (target_width // image_size)) + 1) * image_size
+        )
+        processed_images.append(resized_img.crop(box))
+    if use_thumbnail and len(processed_images) != 1:
+        thumbnail_img = image.resize((image_size, image_size))
+        processed_images.append(thumbnail_img)
+    return processed_images
+
+def load_image(image_file, input_size=448, max_num=12):
+    image = Image.open(image_file).convert('RGB')
+    transform = build_transform(input_size)
+    images = dynamic_preprocess(image, image_size=input_size, use_thumbnail=True, max_num=max_num)
+    return torch.stack([transform(img) for img in images])
+
+
+model_path = 'OpenGVLab/InternVL2_5-4B'
+model = AutoModel.from_pretrained(
+    model_path,
+    torch_dtype=torch.bfloat16,
+    low_cpu_mem_usage=True,
+    use_flash_attn=True,
+    cache_dir="./",
+    trust_remote_code=True
+).eval().cuda()
+tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True, use_fast=False)
+
+
+question = (
+'''<image>\n
+1. Identify the type of fruit or crop shown in the image.  
+2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  
+3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  
+4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).
+Please respond in the following format:
+Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  Recommendation: [keep for further growth /  pick it / try to recover it / discard it]  Consumer Score: [1-100]
+ '''
+)
+
+
+generation_config = dict(max_new_tokens=1024, do_sample=True)
+
+
+root_folder = "../data"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path): continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path): continue
+
+        image_files = [f for f in os.listdir(subfolder_path)
+                       if f.lower().endswith(('.jpg', '.jpeg', '.png', '.bmp'))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+        
+        with open(output_file, "w", encoding="utf-8") as out_file:
+            for image_file in tqdm(image_files, desc=f"{fruit}/{subfolder}"):
+                image_path = os.path.join(subfolder_path, image_file)
+                try:
+                    pixel_values = load_image(image_path).to(torch.bfloat16).cuda()
+                    response = model.chat(tokenizer, pixel_values, question, generation_config)
+
+                    print(f"{image_file} ✅ -> {response}")
+                    out_file.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    out_file.write(f"🖼️ Image Name: {image_file}\n")
+                    out_file.write(f"📝 Answer:\n{response.strip()}\n")
+                    out_file.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{image_file}: {e}")
+                    out_file.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    out_file.write(f"🖼️ Image Name: {image_file}\n")
+                    out_file.write(f"❌ ERROR:\n{e}\n")
+                    out_file.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")

script/InterVL2.5-8B-0-shot.py

@@ -0,0 +1,142 @@
+import os
+import torch
+from PIL import Image
+from tqdm import tqdm
+from transformers import AutoModel, AutoTokenizer
+
+from torchvision import transforms as T
+from torchvision.transforms.functional import InterpolationMode
+
+IMAGENET_MEAN = (0.485, 0.456, 0.406)
+IMAGENET_STD = (0.229, 0.224, 0.225)
+
+def build_transform(input_size):
+    return T.Compose([
+        T.Lambda(lambda img: img.convert('RGB') if img.mode != 'RGB' else img),
+        T.Resize((input_size, input_size), interpolation=InterpolationMode.BICUBIC),
+        T.ToTensor(),
+        T.Normalize(mean=IMAGENET_MEAN, std=IMAGENET_STD)
+    ])
+
+def find_closest_aspect_ratio(aspect_ratio, target_ratios, width, height, image_size):
+    best_ratio_diff = float('inf')
+    best_ratio = (1, 1)
+    area = width * height
+    for ratio in target_ratios:
+        target_aspect_ratio = ratio[0] / ratio[1]
+        ratio_diff = abs(aspect_ratio - target_aspect_ratio)
+        if ratio_diff < best_ratio_diff:
+            best_ratio_diff = ratio_diff
+            best_ratio = ratio
+        elif ratio_diff == best_ratio_diff:
+            if area > 0.5 * image_size * image_size * ratio[0] * ratio[1]:
+                best_ratio = ratio
+    return best_ratio
+
+def dynamic_preprocess(image, min_num=1, max_num=12, image_size=448, use_thumbnail=False):
+    orig_width, orig_height = image.size
+    aspect_ratio = orig_width / orig_height
+
+    target_ratios = sorted(
+        [(i, j) for n in range(min_num, max_num + 1) for i in range(1, n + 1) for j in range(1, n + 1)
+         if i * j <= max_num and i * j >= min_num],
+        key=lambda x: x[0] * x[1]
+    )
+
+    target_aspect_ratio = find_closest_aspect_ratio(aspect_ratio, target_ratios, orig_width, orig_height, image_size)
+    target_width = image_size * target_aspect_ratio[0]
+    target_height = image_size * target_aspect_ratio[1]
+    blocks = target_aspect_ratio[0] * target_aspect_ratio[1]
+
+    resized_img = image.resize((target_width, target_height))
+    processed_images = []
+    for i in range(blocks):
+        box = (
+            (i % (target_width // image_size)) * image_size,
+            (i // (target_width // image_size)) * image_size,
+            ((i % (target_width // image_size)) + 1) * image_size,
+            ((i // (target_width // image_size)) + 1) * image_size
+        )
+        split_img = resized_img.crop(box)
+        processed_images.append(split_img)
+
+    if use_thumbnail and len(processed_images) != 1:
+        thumbnail_img = image.resize((image_size, image_size))
+        processed_images.append(thumbnail_img)
+    return processed_images
+
+def load_image(image_file, input_size=448, max_num=12):
+    image = Image.open(image_file).convert('RGB')
+    transform = build_transform(input_size)
+    images = dynamic_preprocess(image, image_size=input_size, use_thumbnail=True, max_num=max_num)
+    pixel_values = [transform(img) for img in images]
+    return torch.stack(pixel_values)
+
+
+model_path = "OpenGVLab/InternVL2_5-8B"
+model = AutoModel.from_pretrained(
+    model_path,
+    torch_dtype=torch.bfloat16,
+    low_cpu_mem_usage=True,
+    use_flash_attn=True,
+    cache_dir="./",
+    trust_remote_code=True
+).eval().cuda()
+tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True, use_fast=False)
+
+generation_config = dict(max_new_tokens=1024, do_sample=True)
+
+
+question = (
+'''<image>\n
+1. Identify the type of fruit or crop shown in the image.  
+2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  
+3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  
+4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).
+Please respond in the following format:
+Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  Recommendation: [keep for further growth / pick it / try to recover it / discard it]  Consumer Score: [1-100]
+ '''
+)
+
+
+root_folder = "../data/"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.jpg', '.jpeg', '.png', '.bmp'))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+       
+        with open(output_file, "w", encoding="utf-8") as out_file:
+            for image_file in tqdm(image_files, desc=f"Processing {fruit}/{subfolder}"):
+                image_path = os.path.join(subfolder_path, image_file)
+                try:
+                    pixel_values = load_image(image_path, max_num=12).to(torch.bfloat16).cuda()
+
+                    response = model.chat(tokenizer, pixel_values, question, generation_config)
+
+                    print(f"{image_file} ✅ -> {response}")
+                    out_file.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    out_file.write(f"🖼️ Image Name: {image_file}\n")
+                    out_file.write(f"📝 Answer:\n{response}\n")
+                    out_file.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{image_file}: {e}")
+                    out_file.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    out_file.write(f"🖼️ Image Name: {image_file}\n")
+                    out_file.write(f"❌ ERROR:\n{str(e)}\n")
+                    out_file.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")

script/Janus-Pro-1B-0-shot.py

@@ -0,0 +1,98 @@
+import os
+import torch
+from transformers import AutoModelForCausalLM
+from janus.models import MultiModalityCausalLM, VLChatProcessor
+from janus.utils.io import load_pil_images
+from tqdm import tqdm
+
+#https://github.com/deepseek-ai/Janus
+
+model_path = "deepseek-ai/Janus-1.3B"
+vl_chat_processor: VLChatProcessor = VLChatProcessor.from_pretrained(model_path, cache_dir=".")
+tokenizer = vl_chat_processor.tokenizer
+vl_gpt: MultiModalityCausalLM = AutoModelForCausalLM.from_pretrained(
+    model_path, cache_dir=".", trust_remote_code=True
+).to(torch.bfloat16).cuda().eval()
+
+question = (
+'''
+1. Identify the type of fruit or crop shown in the image.  
+2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  
+3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  
+4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).
+Please respond in the following format:
+Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  Recommendation: [keep for further growth / try to recover it / discard it]  Consumer Score: [1-100]
+ '''
+)
+
+root_folder = "../data/"
+output_root = "result-1B"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.png', '.jpg', '.jpeg', '.bmp'))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+       
+        with open(output_file, "w", encoding="utf-8") as fout:
+            for image_file in tqdm(image_files, desc=f"{fruit}/{subfolder}"):
+                try:
+                    image_path = os.path.join(subfolder_path, image_file)
+
+                   
+                    conversation = [
+                        {
+                            "role": "<|User|>",
+                            "content": f"<image_placeholder>\n{question}",
+                            "images": [image_path],
+                        },
+                        {"role": "<|Assistant|>", "content": ""},
+                    ]
+
+                   
+                    pil_images = load_pil_images(conversation)
+                    prepare_inputs = vl_chat_processor(
+                        conversations=conversation,
+                        images=pil_images,
+                        force_batchify=True
+                    ).to(vl_gpt.device)
+
+                    
+                    inputs_embeds = vl_gpt.prepare_inputs_embeds(**prepare_inputs)
+
+                
+                    outputs = vl_gpt.language_model.generate(
+                        inputs_embeds=inputs_embeds,
+                        attention_mask=prepare_inputs.attention_mask,
+                        pad_token_id=tokenizer.eos_token_id,
+                        bos_token_id=tokenizer.bos_token_id,
+                        eos_token_id=tokenizer.eos_token_id,
+                        max_new_tokens=512,
+                        do_sample=False,
+                        use_cache=True,
+                    )
+
+                    answer = tokenizer.decode(outputs[0].cpu().tolist(), skip_special_tokens=True)
+                    fout.write(f"{'='*25} IMAGE START {'='*25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"📝 Answer:\n{answer.strip()}\n")
+                    fout.write(f"{'='*25} IMAGE END {'='*25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{image_file}: {e}")
+                    fout.write(f"{'='*25} IMAGE START {'='*25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"❌ ERROR:\n{e}\n")
+                    fout.write(f"{'='*25} IMAGE END {'='*25}\n\n")

script/Janus-Pro-7B-0-shot.py

@@ -0,0 +1,98 @@
+import os
+import torch
+from transformers import AutoModelForCausalLM
+from janus.models import MultiModalityCausalLM, VLChatProcessor
+from janus.utils.io import load_pil_images
+from tqdm import tqdm
+
+#https://github.com/deepseek-ai/Janus
+
+
+model_path = "deepseek-ai/Janus-Pro-7B"
+vl_chat_processor: VLChatProcessor = VLChatProcessor.from_pretrained(model_path, cache_dir=".")
+tokenizer = vl_chat_processor.tokenizer
+vl_gpt: MultiModalityCausalLM = AutoModelForCausalLM.from_pretrained(
+    model_path, cache_dir=".", trust_remote_code=True
+).to(torch.bfloat16).cuda().eval()
+
+
+question = (
+'''
+1. Identify the type of fruit or crop shown in the image.  
+2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  
+3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  
+4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).
+Please respond in the following format:
+Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  Recommendation: [keep for further growth / pick it /  try to recover it / discard it]  Consumer Score: [1-100]
+ '''
+)
+
+root_folder = "../data"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.png', '.jpg', '.jpeg', '.bmp'))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+        
+        with open(output_file, "w", encoding="utf-8") as fout:
+            for image_file in tqdm(image_files, desc=f"{fruit}/{subfolder}"):
+                try:
+                    image_path = os.path.join(subfolder_path, image_file)
+
+                   
+                    conversation = [
+                        {
+                            "role": "<|User|>",
+                            "content": f"<image_placeholder>\n{question}",
+                            "images": [image_path],
+                        },
+                        {"role": "<|Assistant|>", "content": ""},
+                    ]
+
+                    pil_images = load_pil_images(conversation)
+                    prepare_inputs = vl_chat_processor(
+                        conversations=conversation,
+                        images=pil_images,
+                        force_batchify=True
+                    ).to(vl_gpt.device)
+
+                   
+                    inputs_embeds = vl_gpt.prepare_inputs_embeds(**prepare_inputs)
+
+                    outputs = vl_gpt.language_model.generate(
+                        inputs_embeds=inputs_embeds,
+                        attention_mask=prepare_inputs.attention_mask,
+                        pad_token_id=tokenizer.eos_token_id,
+                        bos_token_id=tokenizer.bos_token_id,
+                        eos_token_id=tokenizer.eos_token_id,
+                        max_new_tokens=512,
+                        do_sample=False,
+                        use_cache=True,
+                    )
+
+                    answer = tokenizer.decode(outputs[0].cpu().tolist(), skip_special_tokens=True)
+                    fout.write(f"{'='*25} IMAGE START {'='*25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"📝 Answer:\n{answer.strip()}\n")
+                    fout.write(f"{'='*25} IMAGE END {'='*25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{image_file}: {e}")
+                    fout.write(f"{'='*25} IMAGE START {'='*25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"❌ ERROR: {e}\n")
+                    fout.write(f"{'='*25} IMAGE END {'='*25}\n\n")

script/Janus-Pro-7B-1-shot.py

@@ -0,0 +1,181 @@
+# -*- coding: utf-8 -*-
+
+
+import os
+import gc
+import random
+import torch
+from tqdm import tqdm
+from transformers import AutoModelForCausalLM
+from janus.models import MultiModalityCausalLM, VLChatProcessor
+from janus.utils.io import load_pil_images  
+
+
+chinese_to_english = {
+    'strawberry', 'tomato', 'guava', 'dragon fruit',
+    'orange',   'pear',     'lychee',   'mango',
+    'kiwi',    'papaya', 'apple',    'grape',
+    'pomegranate','peach', 'banana',   'pomelo'
+}
+
+stage_to_english = {
+    'unripe', 'mature', 'pest-damaged', 'rotten'
+}
+
+recommendation_map = {
+    'unripe': 'keep for further growth',
+    'mature': 'picking it',
+    'pest-damaged': 'try to recover it',
+    'rotten': 'discard it'
+}
+
+score_map = {
+    'unripe': 30, 'mature': 85, 'pest-damaged': 20, 'rotten': 5
+}
+
+
+model_path  = "deepseek-ai/Janus-Pro-7B"
+root_folder = "../data"            
+output_root = "result_1shot"
+os.makedirs(output_root, exist_ok=True)
+
+print("🚀 Loading Janus‑Pro‑7B ...")
+vl_chat_processor: VLChatProcessor = VLChatProcessor.from_pretrained(model_path, cache_dir=".")
+tokenizer = vl_chat_processor.tokenizer
+vl_gpt: MultiModalityCausalLM = AutoModelForCausalLM.from_pretrained(
+    model_path, cache_dir=".", trust_remote_code=True
+)
+vl_gpt = vl_gpt.to(torch.bfloat16).cuda().eval()
+print("✅ Model ready.\n")
+
+
+question = (
+    "1. Identify the type of fruit or crop shown in the image.  \n"
+    "2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  \n"
+    "3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  \n"
+    "4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).\n"
+    "Please respond in the following format:\n"
+    "Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  "
+    "Recommendation: [keep for further growth / pick it /  try to recover it / discard it]  Consumer Score: [1-100]"
+)
+
+
+
+def build_stage_example(fruit_dir: str, fruit_cn: str, stage_cn: str):
+   
+    stage_path = os.path.join(fruit_dir, stage_cn)
+    imgs = [f for f in os.listdir(stage_path)
+            if f.lower().endswith((".png", ".jpg", ".jpeg", ".bmp"))]
+    if not imgs:
+        return []
+
+    img_path = os.path.join(stage_path, random.choice(imgs))
+
+    fruit_en = chinese_to_english.get(fruit_cn, fruit_cn)
+    stage_en  = stage_to_english[stage_cn]
+
+    assistant_reply = (
+        f"Type: {fruit_en}  Growth Stage: {stage_en}  "
+        f"Recommendation: {recommendation_map[stage_en]}  "
+        f"Consumer Score: {score_map[stage_en]}"
+    )
+
+  
+    return [
+        {
+            "role": "<|User|>",
+            "content": f"<image_placeholder>\n{question}",
+            "images": [img_path],
+        },
+        {
+            "role": "<|Assistant|>",
+            "content": assistant_reply,
+        },
+    ]
+
+for fruit_cn in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit_cn)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for stage_cn in os.listdir(fruit_path):
+        stage_path = os.path.join(fruit_path, stage_cn)
+        if not os.path.isdir(stage_path):
+            continue
+
+        img_files = [f for f in os.listdir(stage_path)
+                     if f.lower().endswith((".png", ".jpg", ".jpeg", ".bmp"))]
+
+
+       
+        example_1shot = build_stage_example(fruit_path, fruit_cn, stage_cn)
+        if not example_1shot:
+           
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit_cn}_{stage_cn}.txt")
+        with open(output_file, "w", encoding="utf-8") as fout:
+            for img_name in tqdm(img_files, desc=f"{fruit_cn}/{stage_cn}"):
+                img_path = os.path.join(stage_path, img_name)
+
+               
+                if any(img_path in m.get("images", []) for m in example_1shot):
+                    continue
+
+                conversation = example_1shot + [
+                    {
+                        "role": "<|User|>",
+                        "content": f"<image_placeholder>\n{question}",
+                        "images": [img_path],
+                    },
+                    {"role": "<|Assistant|>", "content": ""},
+                ]
+
+                try:
+                    with torch.no_grad():
+                        pil_imgs = load_pil_images(conversation)
+                        prep_in  = vl_chat_processor(
+                            conversations=conversation,
+                            images=pil_imgs,
+                            force_batchify=True
+                        ).to(vl_gpt.device)
+                        
+                        seq_len = len(prep_in.input_ids[0])
+                        
+
+                        tail_text = tokenizer.decode(prep_in.input_ids[0][-120:])
+                       
+                        del pil_imgs
+
+                        embeds = vl_gpt.prepare_inputs_embeds(**prep_in)
+                        outputs = vl_gpt.language_model.generate(
+                            inputs_embeds=embeds,
+                            attention_mask=prep_in.attention_mask,
+                            pad_token_id=tokenizer.eos_token_id,
+                            bos_token_id=tokenizer.bos_token_id,
+                            eos_token_id=tokenizer.eos_token_id,
+                            max_new_tokens=256,
+                            do_sample=False,
+                            use_cache=True,
+                        )
+                        answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+               
+                    print(f"\n🖼️ {img_name}\n{answer.strip()}")
+                    fout.write("="*25 + " IMAGE START " + "="*25 + "\n")
+                    fout.write(f"🖼️ Image Name: {img_name}\n")
+                    fout.write(f"📝 Answer:\n{answer.strip()}\n")
+                    fout.write("="*25 + " IMAGE END " + "="*25 + "\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit_cn}/{stage_cn}/{img_name}: {e}")
+                    fout.write("="*25 + " IMAGE START " + "="*25 + "\n")
+                    fout.write(f"🖼️ Image Name: {img_name}\n")
+                    fout.write(f"❌ ERROR: {e}\n")
+                    fout.write("="*25 + " IMAGE END " + "="*25 + "\n\n")
+
+                finally:
+                    for var in ("prep_in", "embeds", "outputs"):
+                        if var in locals():
+                            del locals()[var]
+                    torch.cuda.empty_cache()

script/Mantis-8B-Idefics-0-shot.py

@@ -0,0 +1,93 @@
+import os
+import torch
+from PIL import Image
+from tqdm import tqdm
+from transformers import AutoProcessor, AutoModelForVision2Seq
+
+
+model_path = "TIGER-Lab/Mantis-8B-Idefics2"
+processor = AutoProcessor.from_pretrained(model_path, cache_dir=".")
+model = AutoModelForVision2Seq.from_pretrained(
+    model_path,
+    cache_dir=".",
+    device_map="auto"
+)
+model.eval()
+
+
+generation_kwargs = {
+    "max_new_tokens": 1024,
+    "num_beams": 1,
+    "do_sample": False
+}
+
+
+query = (
+'''
+1. Identify the type of fruit or crop shown in the image.  
+2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  
+3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  
+4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).
+Please respond in the following format:
+Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  Recommendation: [keep for further growth / pick it / try to recover it / discard it]  Consumer Score: [1-100]
+ '''
+)
+root_folder = "../data/"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith((".jpg", ".jpeg", ".png", ".bmp"))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+     
+        with open(output_file, "w", encoding="utf-8") as fout:
+            for image_file in tqdm(image_files, desc=f"Processing {fruit}/{subfolder}"):
+                try:
+                    image_path = os.path.join(subfolder_path, image_file)
+                    image = Image.open(image_path).convert("RGB")
+
+                    messages = [
+                        {
+                            "role": "user",
+                            "content": [
+                                {"type": "image"},
+                                {"type": "text", "text": query},
+                            ]
+                        }
+                    ]
+
+                  
+                    prompt = processor.apply_chat_template(messages, add_generation_prompt=True)
+                    inputs = processor(text=prompt, images=[image], return_tensors="pt")
+                    inputs = {k: v.to(model.device) for k, v in inputs.items()}
+
+                    with torch.no_grad():
+                        generated_ids = model.generate(**inputs, **generation_kwargs)
+                        response = processor.batch_decode(
+                            generated_ids[:, inputs["input_ids"].shape[1]:],
+                            skip_special_tokens=True
+                        )[0].strip()
+
+                    fout.write(f"{'='*25} IMAGE START {'='*25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"📝 Answer:\n{response}\n")
+                    fout.write(f"{'='*25} IMAGE END {'='*25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{image_file}: {e}")
+                    fout.write(f"{'='*25} IMAGE START {'='*25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"❌ ERROR: {e}\n")
+                    fout.write(f"{'='*25} IMAGE END {'='*25}\n\n")

script/Mantis-8B-siglip-0-shot.py

@@ -0,0 +1,77 @@
+import os
+from PIL import Image
+from mantis.models.mllava import chat_mllava, MLlavaProcessor, LlavaForConditionalGeneration
+
+#https://huggingface.co/TIGER-Lab/Mantis-8B-siglip-llama3
+
+
+model_path = "TIGER-Lab/Mantis-8B-siglip-llama3"
+processor = MLlavaProcessor.from_pretrained(model_path, cache_dir=".")
+model = LlavaForConditionalGeneration.from_pretrained(
+    model_path,
+    cache_dir=".",
+    device_map="cuda",
+    torch_dtype="bfloat16",
+    attn_implementation=None  # or "flash_attention_2"
+)
+
+
+generation_kwargs = {
+    "max_new_tokens": 1024,
+    "num_beams": 1,
+    "do_sample": False
+}
+
+question = (
+'''
+1. Identify the type of fruit or crop shown in the image.  
+2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  
+3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  
+4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).
+Please respond in the following format:
+Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  Recommendation: [keep for further growth / pick it /try to recover it / discard it]  Consumer Score: [1-100]
+ '''
+)
+
+root_folder = "../data/"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.jpg', '.jpeg', '.png', '.bmp'))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+     
+        with open(output_file, "w", encoding="utf-8") as fout:
+            for image_file in image_files:
+                image_path = os.path.join(subfolder_path, image_file)
+                try:
+                    image = Image.open(image_path).convert("RGB")
+
+                    response, history = chat_mllava(
+                        question, [image], model, processor, **generation_kwargs
+                    )
+
+                    print(f"{image_file} ✅ -> {response}")
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"📝 Answer:\n{response.strip()}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{image_file} -> {e}")
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"❌ ERROR:\n{str(e)}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")

script/Mantis-8B-siglip-1-shot.py

@@ -0,0 +1,120 @@
+import os
+import gc,torch
+from PIL import Image
+from mantis.models.mllava import chat_mllava, MLlavaProcessor, LlavaForConditionalGeneration
+
+chinese_to_english = {
+    'strawberry', 'tomato', 'guava', 'dragon fruit',
+    'orange',   'pear',     'lychee',   'mango',
+    'kiwi',    'papaya', 'apple',    'grape',
+    'pomegranate','peach', 'banana',   'pomelo'
+}
+
+stage_to_english = {
+    'unripe', 'mature', 'pest-damaged', 'rotten'
+}
+
+recommendation_map = {
+    'unripe': 'keep for further growth',
+    'mature': 'picking it',
+    'pest-damaged': 'try to recover it',
+    'rotten': 'discard it'
+}
+
+score_map = {
+    'unripe': 30, 'mature': 85, 'pest-damaged': 20, 'rotten': 5
+}
+
+
+model_path = "TIGER-Lab/Mantis-8B-siglip-llama3"
+processor = MLlavaProcessor.from_pretrained(model_path, cache_dir=".")
+model = LlavaForConditionalGeneration.from_pretrained(
+    model_path,
+    cache_dir=".",
+    device_map="cuda",
+    torch_dtype=torch.bfloat16,
+    attn_implementation=None  # or "flash_attention_2"
+)
+
+
+
+generation_kwargs = {
+    "max_new_tokens": 1024,
+    "num_beams": 1,
+    "do_sample": False
+}
+
+base_question = (
+    "1. Identify the type of fruit or crop shown in the image.  \n"
+    "2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  \n"
+    "3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  \n"
+    "4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).\n"
+    "Please respond in the following format:\n"
+    "Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  "
+    "Recommendation: [keep for further growth / try to recover it / discard it]  Consumer Score: [1-100]"
+)
+
+root_folder = "../data/"
+output_root = "result_1shot_mantis"
+os.makedirs(output_root, exist_ok=True)
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    fruit_en = chinese_to_english.get(fruit, fruit)
+
+    for stage_cn in os.listdir(fruit_path):
+        stage_path = os.path.join(fruit_path, stage_cn)
+        if not os.path.isdir(stage_path):
+            continue
+
+        stage_en = stage_to_english.get(stage_cn, 'mature')
+        recomm = recommendation_map.get(stage_en, 'picking it')
+        score = score_map.get(stage_en, 85)
+
+        example_answer = (
+            f"Type: {fruit_en}  Growth Stage: {stage_en}  "
+            f"Recommendation: {recomm}  Consumer Score: {score}"
+        )
+        print(f"🍏 {stage_path} -> -> {example_answer}")
+        image_files = sorted([f for f in os.listdir(stage_path) if f.lower().endswith(('.jpg', '.jpeg', '.png', '.bmp'))])
+       
+        sample_file = image_files[0]
+        sample_path = os.path.join(stage_path, sample_file)
+        sample_image = Image.open(sample_path).convert("RGB")
+
+        output_file = os.path.join(output_root, f"{fruit}_{stage_cn}.txt")
+        
+        with open(output_file, "w", encoding="utf-8") as fout:
+            for image_file in image_files:
+                if image_file == sample_file:
+                    continue
+
+                image_path = os.path.join(stage_path, image_file)
+                try:
+                    image = Image.open(image_path).convert("RGB")
+
+                    full_prompt = (
+                        f"<image>\n1. Identify the fruit.\n2. Current growth stage.\n3. Recommended action.\n4. Consumer willingness score.\n"
+                        f"{example_answer}\n<image>\n{base_question}"
+                    )
+
+                    response, _ = chat_mllava(full_prompt, [sample_image, image], model, processor, **generation_kwargs)
+
+                    print(f"✅ {image_file} => {response.splitlines()[0] if response else 'N/A'}")
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"📝 Answer:\n{response.strip()}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{stage_cn}/{image_file} -> {e}")
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {image_file}\n")
+                    fout.write(f"❌ ERROR:\n{str(e)}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+                finally:
+                    gc.collect()

script/MiniCPM-0-2.6-8B-0-shot.py

@@ -0,0 +1,83 @@
+import os
+import torch
+from PIL import Image
+from transformers import AutoModel, AutoTokenizer
+from tqdm import tqdm
+
+MODEL_PATH = 'openbmb/MiniCPM-o-2_6'
+
+
+model = AutoModel.from_pretrained(
+    MODEL_PATH,
+    trust_remote_code=True,
+    cache_dir=".",
+    attn_implementation='sdpa',
+    torch_dtype=torch.bfloat16,
+    init_vision=True,
+    init_audio=True,
+    init_tts=True
+).eval().cuda()
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH, cache_dir=".",trust_remote_code=True)
+
+
+model.init_tts()
+
+
+question = (
+    '''You are an agricultural expert. Analyze the image and answer the following questions:
+
+1. Identify the type of fruit or crop shown in the image.  
+2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  
+3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  
+4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).
+
+Please respond in the following format, and do not include explanations:
+
+- Type: [Fruit/Crop Name]  
+- Growth Stage: [unripe / mature / pest-damaged / rotten]  
+- Recommendation: [keep for further growth / pick it / try to recover it / discard it]  
+- Consumer Score: [1-100]
+'''
+)
+
+image_root = "../data"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(image_root):
+    fruit_path = os.path.join(image_root, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.jpg', '.jpeg', '.png', '.bmp', '.webp'))]
+
+        with open(output_file, "w", encoding="utf-8") as out_file:
+            for filename in tqdm(image_files, desc=f"Processing {fruit}/{subfolder}"):
+                image_path = os.path.join(subfolder_path, filename)
+                try:
+                    image = Image.open(image_path).convert('RGB')
+                    msgs = [{'role': 'user', 'content': [image, question]}]
+
+                    response = model.chat(
+                        msgs=msgs,
+                        tokenizer=tokenizer
+                    )
+
+                    out_file.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    out_file.write(f"🖼️ Image Name: {filename}\n\n")
+                    out_file.write("📝 Answer:\n" + response.strip() + "\n")
+                    out_file.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+                except Exception as e:
+                    out_file.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    out_file.write(f"🖼️ Image Name: {filename}\n")
+                    out_file.write("❌ ERROR:\n" + str(e) + "\n")
+                    out_file.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+                    print(f"[ERROR] {fruit}/{subfolder}/{filename}: {e}")

script/Qwen2.5-VL-3B-0-shot.py

@@ -0,0 +1,97 @@
+import os
+import torch
+from tqdm import tqdm
+from transformers import Qwen2_5_VLForConditionalGeneration, AutoTokenizer, AutoProcessor
+from qwen_vl_utils import process_vision_info
+from PIL import Image
+
+
+model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
+    "Qwen/Qwen2.5-VL-3B-Instruct", torch_dtype=torch.bfloat16, device_map="auto", cache_dir="."
+).eval()
+
+processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-3B-Instruct", cache_dir=".")
+
+
+question = '''
+You are an agricultural expert. Analyze the image and answer the following questions:
+1. Identify the type of fruit or crop shown in the image.  
+2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  
+3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  
+4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).
+
+Please respond in the following format, and do not include explanations:
+
+- Type: [Fruit/Crop Name]  
+- Growth Stage: [unripe / mature / pest-damaged / rotten]  
+- Recommendation: [keep for further growth / try to recover it / discard it]  
+- Consumer Score: [1-100]
+'''
+
+
+root_folder = "../data"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path): continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path): continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.jpg', '.jpeg', '.png', '.bmp'))]
+        if not image_files: continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+       
+        with open(output_file, "w", encoding="utf-8") as fout:
+            for img_file in tqdm(image_files, desc=f"{fruit}/{subfolder}"):
+                try:
+                    image_path = os.path.join(subfolder_path, img_file)
+                    messages = [
+                        {
+                            "role": "user",
+                            "content": [
+                                {"type": "image", "image": image_path},
+                                {"type": "text", "text": question},
+                            ]
+                        }
+                    ]
+
+          
+                    text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+                    image_inputs, video_inputs = process_vision_info(messages)
+                    inputs = processor(
+                        text=[text],
+                        images=image_inputs,
+                        videos=video_inputs,
+                        padding=True,
+                        return_tensors="pt",
+                    ).to("cuda")
+
+              
+                    with torch.no_grad():
+                        generated_ids = model.generate(**inputs, max_new_tokens=128)
+                        generated_ids_trimmed = [
+                            out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
+                        ]
+                        output_text = processor.batch_decode(
+                            generated_ids_trimmed,
+                            skip_special_tokens=True,
+                            clean_up_tokenization_spaces=False
+                        )[0].strip()
+
+       
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {img_file}\n")
+                    fout.write(f"📝 Answer:\n{output_text}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{img_file}: {e}")
+                    fout.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    fout.write(f"🖼️ Image Name: {img_file}\n")
+                    fout.write(f"❌ ERROR: {e}\n")
+                    fout.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")

script/Qwen2.5-VL-7B-0-shot.py

@@ -0,0 +1,95 @@
+import os
+from transformers import Qwen2_5_VLForConditionalGeneration, AutoProcessor
+from qwen_vl_utils import process_vision_info
+import torch
+
+# https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct
+
+
+model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
+    "Qwen/Qwen2.5-VL-7B-Instruct",
+    cache_dir=".",
+    torch_dtype="auto",
+    device_map="auto"
+)
+processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-7B-Instruct", cache_dir=".")
+
+question_text = (
+    '''
+    "1. Identify the type of fruit or crop shown in the image.  \n"
+    "2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  \n"
+    "3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  \n"
+    "4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).\n"
+    "Please respond in the following format:\n"
+    "Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  "
+    "Recommendation: [keep for further growth / try to recover it / discard it]  Consumer Score: [1-100]"
+''')
+
+root_folder = "../data"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.png', '.jpg', '.jpeg', '.bmp'))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+
+        with open(output_file, "w", encoding="utf-8") as f_out:
+            for image_file in image_files:
+                image_path = os.path.join(subfolder_path, image_file)
+                try:
+                    messages = [
+                        {
+                            "role": "user",
+                            "content": [
+                                {"type": "image", "image": image_path},
+                                {"type": "text", "text": question_text},
+                            ],
+                        }
+                    ]
+
+               
+                    text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+                    image_inputs, video_inputs = process_vision_info(messages)
+                    inputs = processor(
+                        text=[text],
+                        images=image_inputs,
+                        videos=video_inputs,
+                        padding=True,
+                        return_tensors="pt"
+                    ).to("cuda")
+
+            
+                    generated_ids = model.generate(**inputs, max_new_tokens=128)
+                    generated_ids_trimmed = [
+                        out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
+                    ]
+                    output_text = processor.batch_decode(
+                        generated_ids_trimmed,
+                        skip_special_tokens=True,
+                        clean_up_tokenization_spaces=False
+                    )[0]
+
+                    print(f"{fruit}/{subfolder}/{image_file} -> {output_text}")
+                    f_out.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    f_out.write(f"🖼️ Image Name: {image_file}\n")
+                    f_out.write(f"📝 Answer:\n{output_text}\n")
+                    f_out.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{image_file} 出错：{e}")
+                    f_out.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    f_out.write(f"🖼️ Image Name: {image_file}\n")
+                    f_out.write(f"❌ ERROR:\n{e}\n")
+                    f_out.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")

script/Qwen2.5-VL-7B-1-shot.py

@@ -0,0 +1,122 @@
+import os
+import torch
+from transformers import Qwen2_5_VLForConditionalGeneration, AutoProcessor
+from qwen_vl_utils import process_vision_info
+from PIL import Image
+
+chinese_to_english = {
+    'strawberry', 'tomato', 'guava', 'dragon fruit',
+    'orange',   'pear',     'lychee',   'mango',
+    'kiwi',    'papaya', 'apple',    'grape',
+    'pomegranate','peach', 'banana',   'pomelo'
+}
+
+stage_to_english = {
+    'unripe', 'mature', 'pest-damaged', 'rotten'
+}
+
+recommendation_map = {
+    'unripe': 'keep for further growth',
+    'mature': 'picking it',
+    'pest-damaged': 'try to recover it',
+    'rotten': 'discard it'
+}
+
+score_map = {
+    'unripe': 30, 'mature': 85, 'pest-damaged': 20, 'rotten': 5
+}
+
+model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
+    "Qwen/Qwen2.5-VL-7B-Instruct",
+    cache_dir=".",
+    torch_dtype=torch.float16,
+    device_map="auto"
+)
+processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-7B-Instruct", cache_dir=".")
+
+base_question = (
+    "1. Identify the type of fruit or crop shown in the image.  \n"
+    "2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  \n"
+    "3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  \n"
+    "4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).\n"
+    "Please respond in the following format:\n"
+    "Type: [Fruit/Crop Name]  Growth Stage: [unripe / mature / pest-damaged / rotten]  "
+    "Recommendation: [keep for further growth /pick it / try to recover it / discard it]  Consumer Score: [1-100]"
+)
+
+root_folder = "../data/"
+output_root = "result_1shot_qwen"
+os.makedirs(output_root, exist_ok=True)
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    fruit_en = chinese_to_english.get(fruit, fruit)
+
+    for stage_cn in os.listdir(fruit_path):
+        stage_path = os.path.join(fruit_path, stage_cn)
+        if not os.path.isdir(stage_path):
+            continue
+
+        stage_en = stage_to_english.get(stage_cn, 'mature')
+        recomm = recommendation_map.get(stage_en, 'picking it')
+        score = score_map.get(stage_en, 85)
+        example_answer = f"Type: {fruit_en}  Growth Stage: {stage_en}  Recommendation: {recomm}  Consumer Score: {score}"
+
+
+        
+        
+        image_files = sorted([f for f in os.listdir(stage_path) if f.lower().endswith(('.png', '.jpg', '.jpeg', '.bmp'))])
+      
+
+        sample_file = image_files[0]
+        sample_path = os.path.join(stage_path, sample_file)
+
+        output_file = os.path.join(output_root, f"{fruit}_{stage_cn}.txt")
+      
+        with open(output_file, "w", encoding="utf-8") as f_out:
+            for image_file in image_files:
+                if image_file == sample_file:
+                    continue
+
+                image_path = os.path.join(stage_path, image_file)
+                try:
+                    messages = [
+                        {
+                            "role": "user",
+                            "content": [
+                                {"type": "image", "image": sample_path},
+                                {"type": "text", "text": "<1-shot example>\n" + example_answer},
+                                {"type": "image", "image": image_path},
+                                {"type": "text", "text": base_question},
+                            ],
+                        }
+                    ]
+
+                    text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+                    image_inputs, video_inputs = process_vision_info(messages)
+                    inputs = processor(
+                        text=[text], images=image_inputs, videos=video_inputs,
+                        padding=True, return_tensors="pt"
+                    ).to(model.device)
+
+                    generated_ids = model.generate(**inputs, max_new_tokens=256)
+                    generated_ids_trimmed = [out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)]
+                    output_text = processor.batch_decode(
+                        generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
+                    )[0]
+
+                    print(f"✅ {image_file} => {output_text.splitlines()[0] if output_text else 'N/A'}")
+                    f_out.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    f_out.write(f"🖼️ Image Name: {image_file}\n")
+                    f_out.write(f"📝 Answer:\n{output_text}\n")
+                    f_out.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{stage_cn}/{image_file} 出错：{e}")
+                    f_out.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    f_out.write(f"🖼️ Image Name: {image_file}\n")
+                    f_out.write(f"❌ ERROR:\n{e}\n")
+                    f_out.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")

script/mPLUG-Owl3-7B-0-shot.py

@@ -0,0 +1,97 @@
+import os
+import torch
+from PIL import Image
+from modelscope import AutoConfig, AutoModel, AutoTokenizer
+
+
+model_path = 'iic/mPLUG-Owl3-7B-241101'
+device = "cuda"
+
+
+config = AutoConfig.from_pretrained(model_path, cache_dir=".", trust_remote_code=True)
+print(config)
+
+
+model = AutoModel.from_pretrained(
+    model_path,
+    attn_implementation='flash_attention_2',
+    cache_dir=".",
+    torch_dtype=torch.bfloat16,
+    trust_remote_code=True
+).eval().cuda()
+
+tokenizer = AutoTokenizer.from_pretrained(model_path)
+processor = model.init_processor(tokenizer)
+
+
+question = """<|image|>
+You are an agricultural expert. Analyze the image and answer the following questions:
+
+1. Identify the type of fruit or crop shown in the image.  
+2. Determine its current growth stage. (Options: unripe, mature, pest-damaged, rotten)  
+3. Recommend the farmer’s next action. (Options: keep for further growth, try to recover it, discard it)  
+4. Evaluate the consumer’s willingness to consume this fruit, from 1 (very unlikely) to 100 (very likely).
+
+Please respond in the following format, and do not include explanations:
+
+- Type: [Fruit/Crop Name]  
+- Growth Stage: [unripe / mature / pest-damaged / rotten]  
+- Recommendation: [keep for further growth / pick it / try to recover it / discard it]  
+- Consumer Score: [1-100]"""
+
+
+root_folder = "../data"
+output_root = "result"
+os.makedirs(output_root, exist_ok=True)
+
+
+for fruit in os.listdir(root_folder):
+    fruit_path = os.path.join(root_folder, fruit)
+    if not os.path.isdir(fruit_path):
+        continue
+
+    for subfolder in os.listdir(fruit_path):
+        subfolder_path = os.path.join(fruit_path, subfolder)
+        if not os.path.isdir(subfolder_path):
+            continue
+
+        image_files = [f for f in os.listdir(subfolder_path) if f.lower().endswith(('.jpg', '.jpeg', '.png', '.bmp'))]
+        if not image_files:
+            continue
+
+        output_file = os.path.join(output_root, f"{fruit}_{subfolder}.txt")
+       
+
+        with open(output_file, "w", encoding="utf-8") as f:
+            for image_file in image_files:
+                image_path = os.path.join(subfolder_path, image_file)
+                try:
+                    image = Image.open(image_path).convert('RGB')
+
+                    messages = [
+                        {"role": "user", "content": question},
+                        {"role": "assistant", "content": ""}
+                    ]
+
+                    inputs = processor(messages, images=[image], videos=None)
+                    inputs.to(device)
+                    inputs.update({
+                        'tokenizer': tokenizer,
+                        'max_new_tokens': 100,
+                        'decode_text': True,
+                    })
+
+                    output = model.generate(**inputs)
+                    print(f"{image_file} ✅ -> {output}")
+                    
+                    f.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    f.write(f"🖼️ Image Name: {image_file}\n")
+                    f.write(f"📝 Answer:\n{output}\n")
+                    f.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")
+
+                except Exception as e:
+                    print(f"[ERROR] {fruit}/{subfolder}/{image_file}: {e}")
+                    f.write(f"{'=' * 25} IMAGE START {'=' * 25}\n")
+                    f.write(f"🖼️ Image Name: {image_file}\n")
+                    f.write(f"❌ ERROR:\n{e}\n")
+                    f.write(f"{'=' * 25} IMAGE END {'=' * 25}\n\n")