File size: 4,266 Bytes
a567fa4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 |
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.
<!--
./gen_html_table.py --config 'Detectron1-Comparisons/*.yaml' --name "Faster R-CNN" "Keypoint R-CNN" "Mask R-CNN" --fields lr_sched train_speed inference_speed mem box_AP mask_AP keypoint_AP --base-dir ../../../configs/Detectron1-Comparisons
-->
<table><tbody>
<!-- START TABLE -->
<!-- TABLE HEADER -->
<th valign="bottom">Name</th>
<th valign="bottom">lr<br/>sched</th>
<th valign="bottom">train<br/>time<br/>(s/iter)</th>
<th valign="bottom">inference<br/>time<br/>(s/im)</th>
<th valign="bottom">train<br/>mem<br/>(GB)</th>
<th valign="bottom">box<br/>AP</th>
<th valign="bottom">mask<br/>AP</th>
<th valign="bottom">kp.<br/>AP</th>
<th valign="bottom">model id</th>
<th valign="bottom">download</th>
<!-- TABLE BODY -->
<!-- ROW: faster_rcnn_R_50_FPN_noaug_1x -->
<tr><td align="left"><a href="faster_rcnn_R_50_FPN_noaug_1x.yaml">Faster R-CNN</a></td>
<td align="center">1x</td>
<td align="center">0.219</td>
<td align="center">0.038</td>
<td align="center">3.1</td>
<td align="center">36.9</td>
<td align="center"></td>
<td align="center"></td>
<td align="center">137781054</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/Detectron1-Comparisons/faster_rcnn_R_50_FPN_noaug_1x/137781054/model_final_7ab50c.pkl">model</a> | <a href="https://dl.fbaipublicfiles.com/detectron2/Detectron1-Comparisons/faster_rcnn_R_50_FPN_noaug_1x/137781054/metrics.json">metrics</a></td>
</tr>
<!-- ROW: keypoint_rcnn_R_50_FPN_1x -->
<tr><td align="left"><a href="keypoint_rcnn_R_50_FPN_1x.yaml">Keypoint R-CNN</a></td>
<td align="center">1x</td>
<td align="center">0.313</td>
<td align="center">0.071</td>
<td align="center">5.0</td>
<td align="center">53.1</td>
<td align="center"></td>
<td align="center">64.2</td>
<td align="center">137781195</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/Detectron1-Comparisons/keypoint_rcnn_R_50_FPN_1x/137781195/model_final_cce136.pkl">model</a> | <a href="https://dl.fbaipublicfiles.com/detectron2/Detectron1-Comparisons/keypoint_rcnn_R_50_FPN_1x/137781195/metrics.json">metrics</a></td>
</tr>
<!-- ROW: mask_rcnn_R_50_FPN_noaug_1x -->
<tr><td align="left"><a href="mask_rcnn_R_50_FPN_noaug_1x.yaml">Mask R-CNN</a></td>
<td align="center">1x</td>
<td align="center">0.273</td>
<td align="center">0.043</td>
<td align="center">3.4</td>
<td align="center">37.8</td>
<td align="center">34.9</td>
<td align="center"></td>
<td align="center">137781281</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/Detectron1-Comparisons/mask_rcnn_R_50_FPN_noaug_1x/137781281/model_final_62ca52.pkl">model</a> | <a href="https://dl.fbaipublicfiles.com/detectron2/Detectron1-Comparisons/mask_rcnn_R_50_FPN_noaug_1x/137781281/metrics.json">metrics</a></td>
</tr>
</tbody></table>
## 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.
See [this article](https://ppwwyyxx.com/blog/2021/Where-are-Pixels/) for details.
For speed comparison, see [benchmarks](https://detectron2.readthedocs.io/notes/benchmarks.html).
|