Upload 7 files

best_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a10d56e8da9c72b0531088816c554fd4b52646aa96da7ce2d2e8c03ad80526f
+size 44794379

categories.json

@@ -0,0 +1,27 @@
+{
+  "smartphone": {
+    "asaxiy": "https://asaxiy.uz/product/telefony-i-gadzhety/telefony/smartfony",
+    "texnomart": "https://texnomart.uz/ru/katalog/smartfony/",
+    "olcha": "https://olcha.uz/ru/category/telefony-gadzhety-aksessuary/telefony"
+  },
+  "laptop": {
+    "asaxiy": "https://asaxiy.uz/product/kompyutery-i-orgtehnika/noutbuki/noutbuki-2",
+    "texnomart": "https://texnomart.uz/ru/katalog/noutbuki/",
+    "olcha": "https://olcha.uz/ru/category/noutbuki-planshety-kompyutery/noutbuki"
+  },
+  "tv": {
+    "asaxiy": "https://asaxiy.uz/product/televizory-video-i-audio/televizory",
+    "texnomart": "https://texnomart.uz/ru/katalog/televizory/",
+    "olcha": "https://olcha.uz/ru/category/televizory-audio-i-videotekhnika/televizory"
+  },
+  "refrigerator": {
+    "asaxiy": "https://asaxiy.uz/product/bytovaya-tehnika/krupnaya-tehnika-dlya-kuhni/holodilniki",
+    "texnomart": "https://texnomart.uz/ru/katalog/holodilniki/",
+    "olcha": "https://olcha.uz/ru/category/tekhnika-dlya-kukhni/krupnaya-kukhonnaya-tekhnika/kholodilniki-i-morozilnye-kamery-1"
+  },
+  "headphones": {
+    "asaxiy": "https://asaxiy.uz/product/telefony-i-gadzhety/naushniki-i-auditexniki/besprovodniye-naushniki",
+    "texnomart": "https://texnomart.uz/ru/katalog/naushniki",
+    "olcha": "https://olcha.uz/ru/category/telefony-gadzhety-aksessuary/aksessuary/garnitury"
+  }
+}

config.json

@@ -0,0 +1,38 @@
+{
+  "_name_or_path": "google/vit-base-patch16-224-in21k",
+  "architectures": [
+    "ViTForImageClassification"
+  ],
+  "attention_probs_dropout_prob": 0.0,
+  "encoder_stride": 16,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.0,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "headphones",
+    "1": "laptop",
+    "2": "smartphone",
+    "3": "tv",
+    "4": "watch"
+  },
+  "image_size": 224,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "headphones": "0",
+    "laptop": "1",
+    "smartphone": "2",
+    "tv": "3",
+    "watch": "4"
+  },
+  "layer_norm_eps": 1e-12,
+  "model_type": "vit",
+  "num_attention_heads": 12,
+  "num_channels": 3,
+  "num_hidden_layers": 12,
+  "patch_size": 16,
+  "problem_type": "single_label_classification",
+  "qkv_bias": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.46.3"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91a33a308ad342631146ff9e6491891f9042b2cc9142991aac8c712a93c63534
+size 343233204

predict.py

@@ -0,0 +1,42 @@
+"""
+Инференс через дообученный google/vit-base-patch16-224-in21k.
+"""
+
+import sys
+from pathlib import Path
+
+import torch
+from PIL import Image
+from transformers import ViTForImageClassification, ViTImageProcessor
+
+BASE_DIR = Path(__file__).resolve().parent
+MODEL_DIR = BASE_DIR / "models" / "vit-product-classifier"
+DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+
+def predict(img_path: str):
+    if not MODEL_DIR.exists():
+        print(f"Ошибка: папка модели {MODEL_DIR} не найдена. Сначала запусти train.py!")
+        return
+
+    processor = ViTImageProcessor.from_pretrained(str(MODEL_DIR))
+    model = ViTForImageClassification.from_pretrained(str(MODEL_DIR)).to(DEVICE).eval()
+
+    image = Image.open(img_path).convert("RGB")
+    inputs = processor(images=image, return_tensors="pt").to(DEVICE)
+
+    with torch.no_grad():
+        outputs = model(**inputs)
+        probs = torch.softmax(outputs.logits, dim=1)[0]
+        top_prob, top_class_idx = torch.max(probs, 0)
+
+    class_name = model.config.id2label[top_class_idx.item()]
+    print(f"\nИзображение: {img_path}")
+    print(f"Предсказание: {class_name} ({top_prob.item() * 100:.2f}%)")
+
+
+if __name__ == "__main__":
+    if len(sys.argv) > 1:
+        predict(sys.argv[1])
+    else:
+        print("Использование: python predict.py <путь_к_картинке>")

preprocessor_config.json

@@ -0,0 +1,22 @@
+{
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "ViTImageProcessor",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "height": 224,
+    "width": 224
+  }
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec41433b3fa38bffbabc5bba4ca50e3370410041b7fe5280a427a34e47d8be0a
+size 5713