Detectron2 model zoo's experimental settings and a few implementation details are different from Detectron. The differences in implementation details are shared in [Compatibility with Other Libraries](../../docs/notes/compatibility.md). The differences in model zoo's experimental settings include: * Use scale augmentation during training. This improves AP with lower training cost. * Use L1 loss instead of smooth L1 loss for simplicity. This sometimes improves box AP but may affect other AP. * Use `POOLER_SAMPLING_RATIO=0` instead of 2. This does not significantly affect AP. * Use `ROIAlignV2`. This does not significantly affect AP. In this directory, we provide a few configs that __do not__ have the above changes. They mimic Detectron's behavior as close as possible, and provide a fair comparison of accuracy and speed against Detectron.
Name lr
sched
train
time
(s/iter)
inference
time
(s/im)
train
mem
(GB)
box
AP
mask
AP
kp.
AP
model id download
Faster R-CNN 1x 0.219 0.038 3.1 36.9 137781054 model | metrics
Keypoint R-CNN 1x 0.313 0.071 5.0 53.1 64.2 137781195 model | metrics
Mask R-CNN 1x 0.273 0.043 3.4 37.8 34.9 137781281 model | metrics
## Comparisons: * Faster R-CNN: Detectron's AP is 36.7, similar to ours. * Keypoint R-CNN: Detectron's AP is box 53.6, keypoint 64.2. Fixing a Detectron's [bug](https://github.com/facebookresearch/Detectron/issues/459) lead to a drop in box AP, and can be compensated back by some parameter tuning. * Mask R-CNN: Detectron's AP is box 37.7, mask 33.9. We're 1 AP better in mask AP, due to more correct implementation. For speed comparison, see [benchmarks](https://detectron2.readthedocs.io/notes/benchmarks.html).