Repository navigation
Density model: DM-Count on ShanghaiTech A+B
·
6 commits
to v2
since this release
DM-Count density model trained on ShanghaiTech Part A + Part B (VGG19 backbone, MIT-licensed code from cvlab-stonybrook/DM-Count).
How it was trained: 90% of the A and B training photos for training, the other 10% held out to pick the best epoch (epoch 130 of 870). The test photos were not used until the final evaluation.
Test results (mean absolute error, people per photo):
- Part A (packed crowds): 70.1, against 406.9 for the YOLO detector
- Part B (streets): 16.9, against 73.1 for the YOLO detector
SHA-256: d6f72bffcd1ea986011796b43a63e2738fe8c97d19d230ef99286b603ea207b9