adityachaubey commited on
Commit
32da9a2
·
verified ·
1 Parent(s): f1e81a9

Update README.md

Browse files

Added model card .

Files changed (1) hide show
  1. README.md +109 -0
README.md CHANGED
@@ -1,3 +1,112 @@
1
  ---
2
  license: mit
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: mit
3
+ datasets:
4
+ - alkzar90/NIH-Chest-X-ray-dataset
5
+ language:
6
+ - en
7
+ metrics:
8
+ - roc_auc
9
+ base_model:
10
+ - galactixx/Torchvision-DenseNet121-a639ec97
11
+ pipeline_tag: image-classification
12
+ tags:
13
+ - Ai
14
+ - fine-tuned
15
+ - image-classification
16
  ---
17
+ # DenseNet121 Image Classifier
18
+
19
+ This model is a custom image classifier built with **PyTorch**, leveraging a pretrained **DenseNet121** backbone. The model extracts rich visual features from the pretrained weights and uses a custom classification head to predict the target categories.
20
+
21
+ ## Model Details
22
+
23
+ - **Developed by:** [Aditya chaubey]
24
+ - **Model type:** Image Classification (Convolutional Neural Network)
25
+ - **Architecture:** DenseNet121 (Dense Convolutional Network)
26
+ - **Library:** PyTorch
27
+ - **License:** Mit
28
+ - **Pretrained Backbone:** torchvision.models.densenet121
29
+
30
+ ## Intended Uses & Limitations
31
+
32
+ ### Intended Uses
33
+ - Automated classification of images into [14] categories.
34
+ - Fine-tuning or feature extraction for related computer vision tasks.
35
+
36
+ ### Limitations & Bias
37
+ - **Image Resolution:** The backbone expects standard image sizes (typically 322x322 pixels). Performance may degrade on highly distorted or low-resolution images.
38
+ - **Domain Bias:** The model inherits the features learned from ImageNet. It may perform poorly on highly niche domains (e.g., specific medical imaging or satellite data) unless heavily fine-tuned on target data.
39
+
40
+ ## How to Get Started with the Model
41
+
42
+ You can load and run inference with this model directly using standard PyTorch tools.
43
+
44
+ First, ensure you have the `huggingface_hub`, `torch`, and `torchvision` libraries installed:
45
+ ```bash
46
+ pip install torch torchvision huggingface_hub pillow
47
+ ```
48
+
49
+ Use the snippet below to download the model weights and run a prediction:
50
+
51
+ ```python
52
+ import torch
53
+ import torchvision.models as models
54
+ import torchvision.transforms as transforms
55
+ from PIL import Image
56
+ from huggingface_hub import hf_hub_download
57
+
58
+ # 1. Recreate the model architecture
59
+ model = models.densenet121(weights=None) # Initialize architecture
60
+
61
+ # Replace the final classifier head to match your custom number of classes
62
+ # Example: num_classes = 10 (Change this to your actual number of classes)
63
+ num_classes = 10
64
+ model.classifier = torch.nn.Linear(model.classifier.in_features, num_classes)
65
+
66
+ # 2. Download and load the weights from Hugging Face Hub
67
+ # Replace 'your-username/your-model-name' with your repository ID
68
+ checkpoint_path = hf_hub_download(repo_id="adityachaubey/ChestNet", filename="model.pt")
69
+ model.load_state_dict(torch.load(checkpoint_path, map_location=torch.device('cpu')))
70
+ model.eval()
71
+
72
+ # 3. Prepare the input image
73
+ transform = transforms.Compose([
74
+ transforms.Resize(256),
75
+ transforms.CenterCrop(224),
76
+ transforms.ToTensor(),
77
+ transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
78
+ ])
79
+
80
+ image = Image.open("path_to_your_image.jpg").convert("RGB")
81
+ input_tensor = transform(image).unsqueeze(0) # Add batch dimension
82
+
83
+ # 4. Run inference
84
+ with torch.no_grad():
85
+ outputs = model(input_tensor)
86
+ probabilities = torch.nn.functional.softmax(outputs[0], dim=0)
87
+ predicted_class = torch.argmax(probabilities).item()
88
+
89
+ print(f"Predicted Class Index: {predicted_class}")
90
+ ```
91
+
92
+ ## Training Details
93
+
94
+ ### Training Data
95
+ The model was fine-tuned on the [NIH-Chest-X-ray-dataste] dataset, which contains [112,120] images across [14] classes.
96
+
97
+ ### Data Preprocessing
98
+ Images were resized, cropped to 224x224, and normalized using standard ImageNet mean and standard deviation values:
99
+ - **Mean:** `[0.485, 0.456, 0.406]`
100
+ - **Std:** `[0.229, 0.224, 0.225]`
101
+
102
+ ### Training Hyperparameters
103
+ - **Optimizer:** [Adam]
104
+ - **Learning Rate:** [1e-5]
105
+ - **Batch Size:** [e.g., 32]
106
+ - **Epochs:** [10]
107
+
108
+ ## Evaluation Results
109
+
110
+ The model achieved the following performance metrics on the validation/test set:
111
+ - **Accuracy:** [83%]
112
+ - **Loss:** [Insert validation loss]