MODEL_ZOO.md · HRJ/Incremental Object Detection via Meta-Learning

# Detectron2 Model Zoo and Baselines

## Introduction

This file documents a large collection of baselines trained
with detectron2 in Sep-Oct, 2019.
The corresponding configurations for all models can be found under the `configs/` directory.
Unless otherwise noted, the following settings are used for all runs:

#### Common Settings
* All models were trained on [Big Basin](https://engineering.fb.com/data-center-engineering/introducing-big-basin-our-next-generation-ai-hardware/)
  servers with 8 NVIDIA V100 GPUs, with data-parallel sync SGD and a total minibatch size of 16 images.
* All models were trained with CUDA 9.2, cuDNN 7.4.2 or 7.6.3 (the difference in speed is found to be negligible).
* Training curves and other statistics can be found in `metrics` for each model.
* The default settings are __not directly comparable__ with Detectron.
  For example, our default training data augmentation uses scale jittering in addition to horizontal flipping.

For configs that are comparable to Detectron's settings, see
  [Detectron1-Comparisons](configs/Detectron1-Comparisons/) for accuracy comparison,
  and [benchmarks](https://detectron2.readthedocs.io/notes/benchmarks.html)
  for speed comparison.
* Inference speed is measured by `tools/train_net.py --eval-only`, or [inference_on_dataset()](https://detectron2.readthedocs.io/modules/evaluation.html#detectron2.evaluation.inference_on_dataset),
  with batch size 1 in detectron2 directly.
  The actual deployment should in general be faster than the given inference
  speed due to more optimizations.
* Training speed is averaged across the entire training.
	We keep updating the speed with latest version of detectron2/pytorch/etc.,
	so they might be different from the `metrics` file.
* All COCO models were trained on `train2017` and evaluated on `val2017`.
* For Faster/Mask R-CNN, we provide baselines based on __3 different backbone combinations__:
  * __FPN__: Use a ResNet+FPN backbone with standard conv and FC heads for mask and box prediction,
    respectively. It obtains the best
    speed/accuracy tradeoff, but the other two are still useful for research.
  * __C4__: Use a ResNet conv4 backbone with conv5 head. The original baseline in the Faster R-CNN paper.
  * __DC5__ (Dilated-C5): Use a ResNet conv5 backbone with dilations in conv5, and standard conv and FC heads
    for mask and box prediction, respectively.
    This is used by the Deformable ConvNet paper.
* Most models are trained with the 3x schedule (~37 COCO epochs).
  Although 1x models are heavily under-trained, we provide some ResNet-50 models with the 1x (~12 COCO epochs)
  training schedule for comparison when doing quick research iteration.
* The *model id* column is provided for ease of reference.
  To check downloaded file integrity, any model on this page contains its md5 prefix in its file name.
	Each model also comes with a metrics file with all the training statistics and evaluation curves.

#### ImageNet Pretrained Models

We provide backbone models pretrained on ImageNet-1k dataset.
These models are __different__ from those provided in Detectron: we do not fuse BatchNorm into an affine layer.
* [R-50.pkl](https://dl.fbaipublicfiles.com/detectron2/ImageNetPretrained/MSRA/R-50.pkl): converted copy of MSRA's original ResNet-50 model
* [R-101.pkl](https://dl.fbaipublicfiles.com/detectron2/ImageNetPretrained/MSRA/R-101.pkl): converted copy of MSRA's original ResNet-101 model
* [X-101-32x8d.pkl](https://dl.fbaipublicfiles.com/detectron2/ImageNetPretrained/FAIR/X-101-32x8d.pkl): ResNeXt-101-32x8d model trained with Caffe2 at FB

Pretrained models in Detectron's format can still be used. For example:
* [X-152-32x8d-IN5k.pkl](https://dl.fbaipublicfiles.com/detectron/ImageNetPretrained/25093814/X-152-32x8d-IN5k.pkl):
  ResNeXt-152-32x8d model trained on ImageNet-5k with Caffe2 at FB (see ResNeXt paper for details on ImageNet-5k).
* [R-50-GN.pkl](https://dl.fbaipublicfiles.com/detectron/ImageNetPretrained/47261647/R-50-GN.pkl):
  ResNet-50 with Group Normalization.
* [R-101-GN.pkl](https://dl.fbaipublicfiles.com/detectron/ImageNetPretrained/47592356/R-101-GN.pkl):
  ResNet-101 with Group Normalization.

#### License

All models available for download through this document are licensed under the
[Creative Commons Attribution-ShareAlike 3.0 license](https://creativecommons.org/licenses/by-sa/3.0/).

### COCO Object Detection Baselines

#### Faster R-CNN:
<!--
(fb only) To update the table in vim:
1. Remove the old table: d{
2. Copy the below command to the place of the table
3. :.!bash

./gen_html_table.py --config 'COCO-Detection/faster*50*'{1x,3x}'*' 'COCO-Detection/faster*101*' --name R50-C4 R50-DC5 R50-FPN R50-C4 R50-DC5 R50-FPN R101-C4 R101-DC5 R101-FPN X101-FPN --fields lr_sched train_speed inference_speed mem box_AP
-->

<table><tbody>


<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">model id</th>
<th valign="bottom">download</th>


 <tr><td align="left"><a href="configs/COCO-Detection/warp_faster_rcnn_R_50_C4_1x.yaml">R50-C4</a></td>
<td align="center">1x</td>
<td align="center">0.551</td>
<td align="center">0.110</td>
<td align="center">4.8</td>
<td align="center">35.7</td>
<td align="center">137257644</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_C4_1x/137257644/model_final_721ade.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_C4_1x/137257644/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-Detection/faster_rcnn_R_50_DC5_1x.yaml">R50-DC5</a></td>
<td align="center">1x</td>
<td align="center">0.380</td>
<td align="center">0.068</td>
<td align="center">5.0</td>
<td align="center">37.3</td>
<td align="center">137847829</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_DC5_1x/137847829/model_final_51d356.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_DC5_1x/137847829/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-Detection/faster_rcnn_R_50_FPN_1x.yaml">R50-FPN</a></td>
<td align="center">1x</td>
<td align="center">0.210</td>
<td align="center">0.055</td>
<td align="center">3.0</td>
<td align="center">37.9</td>
<td align="center">137257794</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_FPN_1x/137257794/model_final_b275ba.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_FPN_1x/137257794/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-Detection/faster_rcnn_R_50_C4_3x.yaml">R50-C4</a></td>
<td align="center">3x</td>
<td align="center">0.543</td>
<td align="center">0.110</td>
<td align="center">4.8</td>
<td align="center">38.4</td>
<td align="center">137849393</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_C4_3x/137849393/model_final_f97cb7.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_C4_3x/137849393/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-Detection/faster_rcnn_R_50_DC5_3x.yaml">R50-DC5</a></td>
<td align="center">3x</td>
<td align="center">0.378</td>
<td align="center">0.073</td>
<td align="center">5.0</td>
<td align="center">39.0</td>
<td align="center">137849425</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_DC5_3x/137849425/model_final_68d202.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_DC5_3x/137849425/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-Detection/faster_rcnn_R_50_FPN_3x.yaml">R50-FPN</a></td>
<td align="center">3x</td>
<td align="center">0.209</td>
<td align="center">0.047</td>
<td align="center">3.0</td>
<td align="center">40.2</td>
<td align="center">137849458</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_FPN_3x/137849458/model_final_280758.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_50_FPN_3x/137849458/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-Detection/faster_rcnn_R_101_C4_3x.yaml">R101-C4</a></td>
<td align="center">3x</td>
<td align="center">0.619</td>
<td align="center">0.149</td>
<td align="center">5.9</td>
<td align="center">41.1</td>
<td align="center">138204752</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_101_C4_3x/138204752/model_final_298dad.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_101_C4_3x/138204752/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-Detection/faster_rcnn_R_101_DC5_3x.yaml">R101-DC5</a></td>
<td align="center">3x</td>
<td align="center">0.452</td>
<td align="center">0.082</td>
<td align="center">6.1</td>
<td align="center">40.6</td>
<td align="center">138204841</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_101_DC5_3x/138204841/model_final_3e0943.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_101_DC5_3x/138204841/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-Detection/faster_rcnn_R_101_FPN_3x.yaml">R101-FPN</a></td>
<td align="center">3x</td>
<td align="center">0.286</td>
<td align="center">0.063</td>
<td align="center">4.1</td>
<td align="center">42.0</td>
<td align="center">137851257</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_101_FPN_3x/137851257/model_final_f6e8b1.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_R_101_FPN_3x/137851257/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-Detection/faster_rcnn_X_101_32x8d_FPN_3x.yaml">X101-FPN</a></td>
<td align="center">3x</td>
<td align="center">0.638</td>
<td align="center">0.120</td>
<td align="center">6.7</td>
<td align="center">43.0</td>
<td align="center">139173657</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_X_101_32x8d_FPN_3x/139173657/model_final_68b088.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-Detection/faster_rcnn_X_101_32x8d_FPN_3x/139173657/metrics.json">metrics</a></td>
</tr>
</tbody></table>

#### RetinaNet:

#### RPN & Fast R-CNN:

### COCO Instance Segmentation Baselines with Mask R-CNN

<table><tbody>


<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">model id</th>
<th valign="bottom">download</th>


 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_R_50_C4_1x.yaml">R50-C4</a></td>
<td align="center">1x</td>
<td align="center">0.584</td>
<td align="center">0.117</td>
<td align="center">5.2</td>
<td align="center">36.8</td>
<td align="center">32.2</td>
<td align="center">137259246</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_C4_1x/137259246/model_final_9243eb.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_C4_1x/137259246/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_R_50_DC5_1x.yaml">R50-DC5</a></td>
<td align="center">1x</td>
<td align="center">0.471</td>
<td align="center">0.074</td>
<td align="center">6.5</td>
<td align="center">38.3</td>
<td align="center">34.2</td>
<td align="center">137260150</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_DC5_1x/137260150/model_final_4f86c3.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_DC5_1x/137260150/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml">R50-FPN</a></td>
<td align="center">1x</td>
<td align="center">0.261</td>
<td align="center">0.053</td>
<td align="center">3.4</td>
<td align="center">38.6</td>
<td align="center">35.2</td>
<td align="center">137260431</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x/137260431/model_final_a54504.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x/137260431/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_R_50_C4_3x.yaml">R50-C4</a></td>
<td align="center">3x</td>
<td align="center">0.575</td>
<td align="center">0.118</td>
<td align="center">5.2</td>
<td align="center">39.8</td>
<td align="center">34.4</td>
<td align="center">137849525</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_C4_3x/137849525/model_final_4ce675.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_C4_3x/137849525/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_R_50_DC5_3x.yaml">R50-DC5</a></td>
<td align="center">3x</td>
<td align="center">0.470</td>
<td align="center">0.075</td>
<td align="center">6.5</td>
<td align="center">40.0</td>
<td align="center">35.9</td>
<td align="center">137849551</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_DC5_3x/137849551/model_final_84107b.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_DC5_3x/137849551/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml">R50-FPN</a></td>
<td align="center">3x</td>
<td align="center">0.261</td>
<td align="center">0.055</td>
<td align="center">3.4</td>
<td align="center">41.0</td>
<td align="center">37.2</td>
<td align="center">137849600</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x/137849600/model_final_f10217.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x/137849600/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_R_101_C4_3x.yaml">R101-C4</a></td>
<td align="center">3x</td>
<td align="center">0.652</td>
<td align="center">0.155</td>
<td align="center">6.3</td>
<td align="center">42.6</td>
<td align="center">36.7</td>
<td align="center">138363239</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_101_C4_3x/138363239/model_final_a2914c.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_101_C4_3x/138363239/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_R_101_DC5_3x.yaml">R101-DC5</a></td>
<td align="center">3x</td>
<td align="center">0.545</td>
<td align="center">0.155</td>
<td align="center">7.6</td>
<td align="center">41.9</td>
<td align="center">37.3</td>
<td align="center">138363294</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_101_DC5_3x/138363294/model_final_0464b7.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_101_DC5_3x/138363294/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_R_101_FPN_3x.yaml">R101-FPN</a></td>
<td align="center">3x</td>
<td align="center">0.340</td>
<td align="center">0.070</td>
<td align="center">4.6</td>
<td align="center">42.9</td>
<td align="center">38.6</td>
<td align="center">138205316</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_101_FPN_3x/138205316/model_final_a3ec72.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_101_FPN_3x/138205316/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/COCO-InstanceSegmentation/mask_rcnn_X_101_32x8d_FPN_3x.yaml">X101-FPN</a></td>
<td align="center">3x</td>
<td align="center">0.690</td>
<td align="center">0.129</td>
<td align="center">7.2</td>
<td align="center">44.3</td>
<td align="center">39.5</td>
<td align="center">139653917</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_X_101_32x8d_FPN_3x/139653917/model_final_2d9806.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_X_101_32x8d_FPN_3x/139653917/metrics.json">metrics</a></td>
</tr>
</tbody></table>

### COCO Person Keypoint Detection Baselines with Keypoint R-CNN

### COCO Panoptic Segmentation Baselines with Panoptic FPN

### LVIS Instance Segmentation Baselines with Mask R-CNN

Mask R-CNN baselines on the [LVIS dataset](https://lvisdataset.org), v0.5.
These baselines are described in Table 3(c) of the [LVIS paper](https://arxiv.org/abs/1908.03195).

NOTE: the 1x schedule here has the same amount of __iterations__ as the COCO 1x baselines.
They are roughly 24 epochs of LVISv0.5 data.
The final results of these configs have large variance across different runs.

### Cityscapes & Pascal VOC Baselines

Simple baselines for
* Mask R-CNN on Cityscapes instance segmentation (initialized from COCO pre-training, then trained on Cityscapes fine annotations only)
* Faster R-CNN on PASCAL VOC object detection (trained on VOC 2007 train+val + VOC 2012 train+val, tested on VOC 2007 using 11-point interpolated AP)

<table><tbody>


<th valign="bottom">Name</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">box<br/>AP50</th>
<th valign="bottom">mask<br/>AP</th>
<th valign="bottom">model id</th>
<th valign="bottom">download</th>


 <tr><td align="left"><a href="configs/Cityscapes/mask_rcnn_R_50_FPN.yaml">R50-FPN, Cityscapes</a></td>
<td align="center">0.240</td>
<td align="center">0.092</td>
<td align="center">4.4</td>
<td align="center"></td>
<td align="center"></td>
<td align="center">36.5</td>
<td align="center">142423278</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/Cityscapes/mask_rcnn_R_50_FPN/142423278/model_final_af9cf5.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/Cityscapes/mask_rcnn_R_50_FPN/142423278/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/PascalVOC-Detection/faster_rcnn_R_50_C4.yaml">R50-C4, VOC</a></td>
<td align="center">0.537</td>
<td align="center">0.086</td>
<td align="center">4.8</td>
<td align="center">51.9</td>
<td align="center">80.3</td>
<td align="center"></td>
<td align="center">142202221</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/PascalVOC-Detection/faster_rcnn_R_50_C4/142202221/model_final_b1acc2.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/PascalVOC-Detection/faster_rcnn_R_50_C4/142202221/metrics.json">metrics</a></td>
</tr>
</tbody></table>

### Other Settings

Ablations for Deformable Conv and Cascade R-CNN:

Ablations for normalization methods:
(Note: The baseline uses `2fc` head while the others use `4conv1fc` head. According to the
[GroupNorm paper](https://arxiv.org/abs/1803.08494), the change in head does not improve the baseline by much)

A few very large models trained for a long time, for demo purposes:

<table><tbody>


<th valign="bottom">Name</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">PQ</th>
<th valign="bottom">model id</th>
<th valign="bottom">download</th>


 <tr><td align="left"><a href="configs/Misc/panoptic_fpn_R_101_dconv_cascade_gn_3x.yaml">Panoptic FPN R101</a></td>
<td align="center">0.123</td>
<td align="center">11.4</td>
<td align="center">47.4</td>
<td align="center">41.3</td>
<td align="center">46.1</td>
<td align="center">139797668</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/Misc/panoptic_fpn_R_101_dconv_cascade_gn_3x/139797668/model_final_be35db.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/Misc/panoptic_fpn_R_101_dconv_cascade_gn_3x/139797668/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left"><a href="configs/Misc/cascade_mask_rcnn_X_152_32x8d_FPN_IN5k_gn_dconv.yaml">Mask R-CNN X152</a></td>
<td align="center">0.281</td>
<td align="center">15.1</td>
<td align="center">50.2</td>
<td align="center">44.0</td>
<td align="center"></td>
<td align="center">18131413</td>
<td align="center"><a href="https://dl.fbaipublicfiles.com/detectron2/Misc/cascade_mask_rcnn_X_152_32x8d_FPN_IN5k_gn_dconv/18131413/model_0039999_e76410.pkl">model</a>&nbsp;|&nbsp;<a href="https://dl.fbaipublicfiles.com/detectron2/Misc/cascade_mask_rcnn_X_152_32x8d_FPN_IN5k_gn_dconv/18131413/metrics.json">metrics</a></td>
</tr>

 <tr><td align="left">above + test-time aug.</td>
<td align="center"></td>
<td align="center"></td>
<td align="center">51.9</td>
<td align="center">45.9</td>
<td align="center"></td>
<td align="center"></td>
<td align="center"></td>
</tr>
</tbody></table>

HRJ/Incremental Object Detection via Meta-Learning

简介

发行版

贡献者

近期动态

HRJ/Incremental Object Detection via Meta-Learning .gitee-modal { width: 500px !important; }

简介

发行版

贡献者

近期动态

搜索帮助

HRJ/Incremental Object Detection via Meta-Learning