Update README.md
Browse files
README.md
CHANGED
@@ -11,14 +11,6 @@ This repository hosts a **fine-tuned ResNet-50 model** for **Vehicle Type Classi
|
|
11 |
- **Fine-tuning Framework:** PyTorch (`torchvision.models.resnet50`)
|
12 |
- **Optimization:** Trained with Adam optimizer and data augmentation for robust performance
|
13 |
|
14 |
-
## Downloading the Model
|
15 |
-
|
16 |
-
You can download the fine-tuned model from the provided link:
|
17 |
-
|
18 |
-
```sh
|
19 |
-
wget <download_link>/fine_tuned_model.zip
|
20 |
-
unzip fine_tuned_model.zip
|
21 |
-
```
|
22 |
|
23 |
## Usage
|
24 |
|
@@ -38,37 +30,54 @@ import torchvision.models as models
|
|
38 |
import torchvision.transforms as transforms
|
39 |
from PIL import Image
|
40 |
|
41 |
-
#
|
42 |
-
|
43 |
-
|
44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
|
46 |
-
# Load fine-tuned weights
|
47 |
-
model.load_state_dict(torch.load("fine_tuned_model/pytorch_model.bin", map_location=torch.device('cpu')))
|
48 |
-
model.eval() # Set to evaluation mode
|
49 |
|
50 |
-
# Load class
|
51 |
with open("fine_tuned_model/classes.txt", "r") as f:
|
52 |
class_names = f.read().splitlines()
|
53 |
|
54 |
-
|
|
|
|
|
|
|
55 |
transform = transforms.Compose([
|
56 |
-
transforms.Resize((224, 224)),
|
57 |
-
transforms.ToTensor(),
|
58 |
-
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
|
59 |
])
|
60 |
|
61 |
-
# Load
|
62 |
-
image_path = "
|
63 |
-
image = Image.open(image_path).convert("RGB")
|
64 |
-
input_tensor = transform(image).unsqueeze(0)
|
|
|
65 |
|
66 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
67 |
with torch.no_grad():
|
68 |
-
outputs =
|
69 |
-
_, predicted_class = torch.max(outputs, 1)
|
70 |
|
|
|
71 |
print(f"Predicted Vehicle Type: {class_names[predicted_class.item()]}")
|
|
|
72 |
```
|
73 |
|
74 |
## Performance Metrics
|
|
|
11 |
- **Fine-tuning Framework:** PyTorch (`torchvision.models.resnet50`)
|
12 |
- **Optimization:** Trained with Adam optimizer and data augmentation for robust performance
|
13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
|
15 |
## Usage
|
16 |
|
|
|
30 |
import torchvision.transforms as transforms
|
31 |
from PIL import Image
|
32 |
|
33 |
+
# Define the model architecture
|
34 |
+
resnet50 = models.resnet50(pretrained=False)
|
35 |
+
|
36 |
+
# Modify the last layer to match the number of classes (11)
|
37 |
+
num_ftrs = resnet50.fc.in_features
|
38 |
+
resnet50.fc = torch.nn.Linear(num_ftrs, 11)
|
39 |
+
|
40 |
+
# Load trained model weights
|
41 |
+
resnet50.load_state_dict(torch.load("fine_tuned_model/pytorch_model.bin"))
|
42 |
+
resnet50.eval() # Set model to evaluation mode
|
43 |
+
|
44 |
+
print("Model loaded successfully!")
|
45 |
|
|
|
|
|
|
|
46 |
|
47 |
+
# Load class names
|
48 |
with open("fine_tuned_model/classes.txt", "r") as f:
|
49 |
class_names = f.read().splitlines()
|
50 |
|
51 |
+
print("Classes:", class_names)
|
52 |
+
|
53 |
+
|
54 |
+
# Define image transformations (same as training)
|
55 |
transform = transforms.Compose([
|
56 |
+
transforms.Resize((224, 224)), # Resize to match ResNet-50 input size
|
57 |
+
transforms.ToTensor(),
|
58 |
+
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]) # Normalization
|
59 |
])
|
60 |
|
61 |
+
# Load the custom image
|
62 |
+
image_path = "/kaggle/input/sample-image-1/pickup_truck_sample_image.jpg" # Change this to your test image path
|
63 |
+
image = Image.open(image_path).convert("RGB") # Open image and convert to RGB
|
64 |
+
input_tensor = transform(image).unsqueeze(0) # Add batch dimension
|
65 |
+
|
66 |
|
67 |
+
|
68 |
+
# Move to GPU if available
|
69 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
70 |
+
resnet50 = resnet50.to(device)
|
71 |
+
input_tensor = input_tensor.to(device)
|
72 |
+
|
73 |
+
# Get predictions
|
74 |
with torch.no_grad():
|
75 |
+
outputs = resnet50(input_tensor)
|
76 |
+
_, predicted_class = torch.max(outputs, 1) # Get the class with highest score
|
77 |
|
78 |
+
# Print the result
|
79 |
print(f"Predicted Vehicle Type: {class_names[predicted_class.item()]}")
|
80 |
+
|
81 |
```
|
82 |
|
83 |
## Performance Metrics
|