Project on Self-Supervised Learning for Prof. Yann LeCun's Deep Learning Course at NYU, Spring 2020

1. Model architecture and training files:

   • Denoising Auto-encoder
   • Road Map Construction

   python road_map_construction.py --EPOCHS 50 --BATCH_SIZE 128
   

2. Sample notebook showing the generated roadmaps on validation data: Link

3. The directory saved_models contains the trained model checkpoints.

4. The code for object detection is inside the directory object_detection.

Requirements

1. PyTorch
2. Torchvision
3. PyTorch Lightning - for simplifying the training, validation and testing interface.