VGG16 30VNFoods
PyTorch VGG16 trained on the processed 30 Vietnamese Foods dataset.
Dataset
Processed dataset from Kaggle: trinhtrantran122/CS231_30VNFoods_processed.
Preprocessing
- Train:
RandomResizedCrop(224, scale=(0.7, 1.0), ratio=(0.75, 1.33)), horizontal flip, color jitter, ImageNet normalization. - Validation/Test: resize to
224 x 224, ImageNet normalization.
Metrics
- Best validation accuracy:
0.8436 - Test accuracy:
0.8278 - Test loss:
0.6497
Files
best_vgg16_30vnfoods.pt: best PyTorch checkpoint.last_vgg16_30vnfoods.pt: final checkpoint.config.json: class names and preprocessing metadata.metrics.json: validation/test metrics.
- Downloads last month
- 34