A research-grade PyTorch framework for robust object recognition under extreme environmental noise. Implements self-supervised Denoising Autoencoders (DAE) with ResNet/ViT architectures on the official CIFAR-10-C benchmark. Includes Grad-CAM interpretability and automated robustness benchmarking.
computer-vision deep-learning grad-cam pytorch explainable-ai denoising-autoencoder vision-transformer cifar-10-c research-benchmark
-
Updated
Mar 16, 2026 - Python