Skip to content
Snippets Groups Projects
model_zoo.md 8.71 KiB
Newer Older
Kai Chen's avatar
Kai Chen committed
# Benchmark and Model Zoo

Kai Chen's avatar
Kai Chen committed
## Mirror sites

We use AWS as the main site to host our model zoo, and maintain a mirror on aliyun.
Kai Chen's avatar
Kai Chen committed
You can replace `https://s3.ap-northeast-2.amazonaws.com/open-mmlab` with `https://open-mmlab.oss-cn-beijing.aliyuncs.com` in model urls.
Kai Chen's avatar
Kai Chen committed

## Common settings

myownskyW7's avatar
myownskyW7 committed
- All FPN baselines and RPN-C4 baselines were trained using 8 GPU with a batch size of 16 (2 images per GPU). Other C4 baselines were trained using 8 GPU with a batch size of 8 (1 image per GPU).
Kai Chen's avatar
Kai Chen committed
- All models were trained on `coco_2017_train`, and tested on the `coco_2017_val`.
- We use distributed training and BN layer stats are fixed.
- We adopt the same training schedules as Detectron. 1x indicates 12 epochs and 2x indicates 24 epochs, which corresponds to slightly less iterations than Detectron and the difference can be ignored.
- All pytorch-style pretrained backbones on ImageNet are from PyTorch model zoo.
- For fair comparison with other codebases, we report the GPU memory as the maximum value of `torch.cuda.max_memory_allocated()` for all 8 GPUs. Note that this value is usually less than what `nvidia-smi` shows.
- We report the inference time as the overall time including data loading, network forwarding and post processing.
Kai Chen's avatar
Kai Chen committed


## Baselines

### RPN

Kai Chen's avatar
Kai Chen committed
Please refer to [RPN](https://github.com/open-mmlab/mmdetection/blob/master/configs/rpn) for details.

Kai Chen's avatar
Kai Chen committed
### Faster R-CNN

Please refer to [Faster R-CNN](https://github.com/open-mmlab/mmdetection/blob/master/configs/faster_rcnn) for details.
Kai Chen's avatar
Kai Chen committed

### Mask R-CNN

Please refer to [Mask R-CNN](https://github.com/open-mmlab/mmdetection/blob/master/configs/mask_rcnn) for details.
Kai Chen's avatar
Kai Chen committed
### Fast R-CNN (with pre-computed proposals)
Please refer to [Fast R-CNN](https://github.com/open-mmlab/mmdetection/blob/master/configs/fast_rcnn) for details.
Kai Chen's avatar
Kai Chen committed
### RetinaNet
Please refer to [RetinaNet](https://github.com/open-mmlab/mmdetection/blob/master/configs/retinanet) for details.
Kai Chen's avatar
Kai Chen committed
### Cascade R-CNN and Cascade Mask R-CNN
Kai Chen's avatar
Kai Chen committed

Please refer to [Cascade R-CNN](https://github.com/open-mmlab/mmdetection/blob/master/configs/cascade_rcnn) for details.
Please refer to [HTC](https://github.com/open-mmlab/mmdetection/blob/master/configs/htc) for details.
Kai Chen's avatar
Kai Chen committed
### SSD

Please refer to [SSD](https://github.com/open-mmlab/mmdetection/blob/master/configs/ssd) for details.
Kai Chen's avatar
Kai Chen committed

Kai Chen's avatar
Kai Chen committed
### Group Normalization (GN)

Please refer to [Group Normalization](https://github.com/open-mmlab/mmdetection/blob/master/configs/gn) for details.
Kai Chen's avatar
Kai Chen committed
### Weight Standardization
Please refer to [Weight Standardization](https://github.com/open-mmlab/mmdetection/blob/master/configs/gn+ws) for details.
Kai Chen's avatar
Kai Chen committed

Kai Chen's avatar
Kai Chen committed
### Deformable Convolution v2
Kai Chen's avatar
Kai Chen committed

Please refer to [Deformable Convolutional Networks](https://github.com/open-mmlab/mmdetection/blob/master/configs/dcn) for details.
Kai Chen's avatar
Kai Chen committed

### CARAFE: Content-Aware ReAssembly of FEatures
Please refer to [CARAFE](https://github.com/open-mmlab/mmdetection/blob/master/configs/carafe) for details.

### Instaboost

Please refer to [Instaboost](https://github.com/open-mmlab/mmdetection/blob/master/configs/instaboost) for details.

Please refer to [Libra R-CNN](https://github.com/open-mmlab/mmdetection/blob/master/configs/libra_rcnn) for details.
Please refer to [Guided Anchoring](https://github.com/open-mmlab/mmdetection/blob/master/configs/guided_anchoring) for details.
Kai Chen's avatar
Kai Chen committed
### FCOS

Please refer to [FCOS](https://github.com/open-mmlab/mmdetection/blob/master/configs/fcos) for details.
Kai Chen's avatar
Kai Chen committed

Tao Kong's avatar
Tao Kong committed
### FoveaBox

Please refer to [FoveaBox](https://github.com/open-mmlab/mmdetection/blob/master/configs/foveabox) for details.
Kai Chen's avatar
Kai Chen committed
### RepPoints

Please refer to [RepPoints](https://github.com/open-mmlab/mmdetection/blob/master/configs/reppoints) for details.
Kai Chen's avatar
Kai Chen committed

### FreeAnchor

Please refer to [FreeAnchor](https://github.com/open-mmlab/mmdetection/blob/master/configs/free_anchor) for details.
Kai Chen's avatar
Kai Chen committed

Kai Chen's avatar
Kai Chen committed
### Grid R-CNN (plus)

Please refer to [Grid R-CNN](https://github.com/open-mmlab/mmdetection/blob/master/configs/grid_rcnn) for details.
Kai Chen's avatar
Kai Chen committed

### GHM

Please refer to [GHM](https://github.com/open-mmlab/mmdetection/blob/master/configs/ghm) for details.
Kai Chen's avatar
Kai Chen committed

### GCNet

Please refer to [GCNet](https://github.com/open-mmlab/mmdetection/blob/master/configs/gcnet) for details.
Kai Chen's avatar
Kai Chen committed

### HRNet
Please refer to [HRNet](https://github.com/open-mmlab/mmdetection/blob/master/configs/hrnet) for details.
Kai Chen's avatar
Kai Chen committed

### Mask Scoring R-CNN

Please refer to [Mask Scoring R-CNN](https://github.com/open-mmlab/mmdetection/blob/master/configs/ms_rcnn) for details.
Kai Chen's avatar
Kai Chen committed

### Train from Scratch

Please refer to [Rethinking ImageNet Pre-training](https://github.com/open-mmlab/mmdetection/blob/master/configs/scratch) for details.
Kai Chen's avatar
Kai Chen committed

Kai Chen's avatar
Kai Chen committed
### NAS-FPN
Please refer to [NAS-FPN](https://github.com/open-mmlab/mmdetection/blob/master/configs/nas_fpn) for details.

### ATSS
Please refer to [ATSS](https://github.com/open-mmlab/mmdetection/blob/master/configs/atss) for details.

Kai Chen's avatar
Kai Chen committed
### Other datasets

We also benchmark some methods on [PASCAL VOC](https://github.com/open-mmlab/mmdetection/blob/master/configs/pascal_voc), [Cityscapes](https://github.com/open-mmlab/mmdetection/blob/master/configs/cityscapes) and [WIDER FACE](https://github.com/open-mmlab/mmdetection/blob/master/configs/wider_face).
Cao Yuhang's avatar
Cao Yuhang committed
## Speed benchmark
We compare the training speed of Mask R-CNN with some other popular frameworks (The data is copied from [detectron2](https://github.com/facebookresearch/detectron2/blob/master/docs/notes/benchmarks.md)).

| Implementation       | Throughput (img/s) |
|----------------------|--------------------|
| [Detectron2](https://github.com/facebookresearch/detectron2) | 61 |
| [MMDetection](https://github.com/open-mmlab/mmdetection) | 60 |
| [maskrcnn-benchmark](https://github.com/facebookresearch/maskrcnn-benchmark/)   | 51 |
| [tensorpack](https://github.com/tensorpack/tensorpack/tree/master/examples/FasterRCNN) | 50 |
| [simpledet](https://github.com/TuSimple/simpledet/) | 39 |
| [Detectron](https://github.com/facebookresearch/Detectron) | 19 |
| [matterport/Mask_RCNN](https://github.com/matterport/Mask_RCNN/) | 14 |

## Comparison with Detectron2
Kai Chen's avatar
Kai Chen committed
We compare mmdetection with [Detectron2](https://github.com/facebookresearch/detectron2.git) in terms of speed and performance.
We use the commit id [185c27e](https://github.com/facebookresearch/detectron2/tree/185c27e4b4d2d4c68b5627b3765420c6d7f5a659)(30/4/2020) of detectron.
For fair comparison, we install and run both frameworks on the same machine.
### Hardware
Kai Chen's avatar
Kai Chen committed

Kai Chen's avatar
Kai Chen committed
- 8 NVIDIA Tesla V100 (32G) GPUs
- Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz
Kai Chen's avatar
Kai Chen committed

### Software environment
- Python 3.7
- PyTorch 1.4
- CUDA 10.1
- CUDNN 7.6.03
- NCCL 2.4.08
### Performance
Kai Chen's avatar
Kai Chen committed
<table border="1">
Kai Chen's avatar
Kai Chen committed
  <tr>
    <th>Type</th>
    <th>Lr schd</th>
    <th>Detectron2</th>
Kai Chen's avatar
Kai Chen committed
    <th>mmdetection</th>
  </tr>
  <tr>
    <td rowspan="2">Faster R-CNN</td>
    <td>1x</td>
    <td>37.9</td>
    <td>38.0</td>
Kai Chen's avatar
Kai Chen committed
  </tr>
  <tr>
    <td>3x</td>
    <td>40.2</td>
Kai Chen's avatar
Kai Chen committed
    <td>-</td>
  </tr>
Kai Chen's avatar
Kai Chen committed
  <tr>
    <td rowspan="2">Mask R-CNN</td>
Kai Chen's avatar
Kai Chen committed
    <td>1x</td>
    <td>38.6 &amp; 35.2</td>
    <td>38.8 &amp; 35.4</td>
Kai Chen's avatar
Kai Chen committed
  </tr>
  <tr>
    <td>3x</td>
    <td>41.0 &amp; 37.2 </td>
Kai Chen's avatar
Kai Chen committed
    <td>-</td>
  </tr>
  <tr>
    <td rowspan="2">Retinanet</td>
Kai Chen's avatar
Kai Chen committed
    <td>1x</td>
    <td>36.5</td>
    <td>37.0</td>
Kai Chen's avatar
Kai Chen committed
  </tr>
  <tr>
    <td>3x</td>
    <td>37.9</td>
Kai Chen's avatar
Kai Chen committed
    <td>-</td>
  </tr>
Kai Chen's avatar
Kai Chen committed
</table>

### Training Speed
The training speed is measure with s/iter. The lower, the better.
Kai Chen's avatar
Kai Chen committed
<table border="1">
Kai Chen's avatar
Kai Chen committed
  <tr>
    <th>Type</th>
Kai Chen's avatar
Kai Chen committed
    <th>Detectron2</th>
    <th>mmdetection</th>
Kai Chen's avatar
Kai Chen committed
  </tr>
  <tr>
    <td>Faster R-CNN</td>
    <td>0.210</td>
    <td>0.216</td>
Kai Chen's avatar
Kai Chen committed
  </tr>
  <tr>
    <td>Mask R-CNN</td>
    <td>0.261</td>
    <td>0.265</td>
Kai Chen's avatar
Kai Chen committed
  </tr>
Kai Chen's avatar
Kai Chen committed
  <tr>
    <td>Retinanet</td>
    <td>0.200</td>
    <td>0.205</td>
Kai Chen's avatar
Kai Chen committed
  </tr>
Kai Chen's avatar
Kai Chen committed
</table>


### Inference Speed

The inference speed is measured with fps (img/s) on a single GPU, the higher, the better.
To be consistent with Detectron2, we report the pure inference speed (without the time of data loading).
For Mask R-CNN, we exclude the time of RLE encoding in post-processing.
Kai Chen's avatar
Kai Chen committed
We also include the officially reported speed in the parentheses, which is slightly higher
than the results tested on our server due to differences of hardwards.
Kai Chen's avatar
Kai Chen committed
<table border="1">
  <tr>
    <th>Type</th>
Kai Chen's avatar
Kai Chen committed
    <th>Detectron2</th>
    <th>mmdetection</th>
  </tr>
  <tr>
    <td>Faster R-CNN</td>
Kai Chen's avatar
Kai Chen committed
    <td>25.6 (26.3)</td>
    <td>22.2</td>
  </tr>
  <tr>
    <td>Mask R-CNN</td>
Kai Chen's avatar
Kai Chen committed
    <td>22.5 (23.3)</td>
    <td>19.6</td>
Kai Chen's avatar
Kai Chen committed
  <tr>
    <td>Retinanet</td>
Kai Chen's avatar
Kai Chen committed
    <td>17.8 (18.2)</td>
    <td>20.6</td>
Kai Chen's avatar
Kai Chen committed
  </tr>
Kai Chen's avatar
Kai Chen committed
### Training memory

Kai Chen's avatar
Kai Chen committed
<table border="1">
    <th>Detectron2</th>
    <th>mmdetection</th>
  </tr>
  <tr>
    <td>Faster R-CNN</td>
    <td>3.0</td>
    <td>Mask R-CNN</td>
    <td>3.4</td>
    <td>3.9</td>
    <td>Retinanet</td>
    <td>3.9</td>