upload 4 files

app.py

@@ -0,0 +1,47 @@
+# app.py
+import gradio as gr
+import torch
+from torchvision import transforms
+from PIL import Image
+import os
+from model import get_mobilenet_model
+
+# 自动检测是否使用GPU
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+# 加载模型
+weights_path = "best_model.pth"
+model, _, _ = get_mobilenet_model(num_classes=16)
+model.load_state_dict(torch.load(weights_path, map_location=device))
+model.to(device)
+model.eval()
+
+# 设置 class names（替换成你自己的类别）
+class_names = [f"Class {i}" for i in range(16)]  # 或者 ['cat', 'dog', ...]
+
+# 图像预处理
+transform = transforms.Compose([
+    transforms.Resize((224, 224)),
+    transforms.ToTensor(),
+    transforms.Normalize(mean=[0.485, 0.456, 0.406],
+                         std=[0.229, 0.224, 0.225]),
+])
+
+def predict(image):
+    image = Image.fromarray(image).convert("RGB")
+    img_tensor = transform(image).unsqueeze(0).to(device)
+    with torch.no_grad():
+        outputs = model(img_tensor)
+        probs = torch.softmax(outputs, dim=1)
+        top_prob, top_class = torch.max(probs, 1)
+        return {class_names[i]: float(probs[0, i]) for i in range(len(class_names))}
+
+demo = gr.Interface(
+    fn=predict,
+    inputs=gr.Image(type="numpy"),
+    outputs=gr.Label(num_top_classes=3),
+    title="MobileNetV3-Large Classifier"
+)
+
+if __name__ == "__main__":
+    demo.launch()

best_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91f42aa4a7919ca1a910ad61d00e2ee65558188639a325a6517407cec717fd4d
+size 35858303

model.py

@@ -0,0 +1,49 @@
+# model.py
+
+import torch
+import torch.nn as nn
+import torch.optim as optim
+from torchvision import models
+from torch.optim import lr_scheduler
+
+def get_mobilenet_model(num_classes=16):
+    """
+    配置 MobileNetV3-Large 模型、优化器和学习率调度器
+    """
+    model = models.mobilenet_v3_large(pretrained=True)
+
+    # 冻结所有层参数
+    for param in model.parameters():
+        param.requires_grad = False
+
+    # 解冻最后三个倒残差块
+    for name, param in model.named_parameters():
+        if 'features.13' in name or 'features.14' in name or 'features.15' in name:
+            param.requires_grad = True
+
+    # 修改分类器结构
+    model.classifier = nn.Sequential(
+        nn.Linear(960, 512),
+        nn.Hardswish(inplace=True),
+        nn.Dropout(0.5),
+        nn.Linear(512, 256),
+        nn.Dropout(0.3),
+        nn.Linear(256, num_classes)
+    )
+
+    # 设置优化器
+    optimizer = optim.AdamW(
+        filter(lambda p: p.requires_grad, model.parameters()),
+        lr=2e-4,
+        weight_decay=5e-5,
+        eps=1e-6
+    )
+
+    # 设置学习率调度器
+    scheduler = lr_scheduler.CosineAnnealingLR(
+        optimizer,
+        T_max=50,
+        eta_min=1e-6
+    )
+
+    return model, optimizer, scheduler

requirements.txt

@@ -0,0 +1,3 @@
+torch
+torchvision
+gradio