Views
No views yet
timm.vit_base_patch16_224, fine-tunedreport.txt and classification_report.txt)1import torch
2import timm
3import cv2
4import numpy as np
5
6# Load model
7model = timm.create_model('vit_base_patch16_224', pretrained=False, num_classes=2)
8model.load_state_dict(torch.load('pytorch_model.bin'))
9model.eval()
10device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
11model.to(device)
12
13# Preprocess image
14img = cv2.imread('path_to_image.jpg')
15img = cv2.resize(img, (224, 224))
16img = img.astype(np.float32) / 255.0
17img = torch.tensor(img).permute(2, 0, 1).unsqueeze(0).to(device)
18
19# Predict
20with torch.no_grad():
21 outputs = model(img)
22 probs = torch.softmax(outputs, dim=1)
23 predicted_idx = outputs.max(1)[1].item()
24 class_names = ['Cancer', 'Normal']
25 confidence = probs[0][predicted_idx].item()
26 print(f'Classified as: {class_names[predicted_idx]}, Confidence: {confidence:.4f}')pytorch_model.bin: Model weightsconfig.json: Model configurationreport.txt: Comprehensive training reportclassification_report.txt: Test set classification metricsconfusion_matrix.png: Confusion matrix plotroc_curve.png: ROC curve with AUCtraining_plots.png: Training loss and validation accuracy plotsreport.txt and classification_report.txt for details on dataset, hyperparameters, and performance.