README.md

Lab 2: Single Object Detection

Implement a single object detector with PyTorch, and train on Tiny-VID dataset. For each classes, use 150 samples for training and the rest for testing.

Prequisity

1. Watch all parts till Object Detection of CS231n.
2. Paper reading: Faster R-CNN, [Optional] YOLO
3. PyTorch Tutorials: https://pytorch.org/tutorials. Only Introduction to PyTorch and Learning PyTorch parts are needed.

Handbook

1. Design a detector, and train & evaluate on Tiny-VID dataset. Complete TvidDataset, Detector, compute_iou and the rest in main.py. The accuracy should be at least 70%.
2. Visualize detection results.
3. Expected completion time: 2~3 days.