mmdeploy/docs/backends/openvino.md
Semyon Bevzyuk ca93007035
[Enhancement]: Support FoveaBox and ATSS from mmdetection for OpenVINO and ONNX Runtime (#160)
* Add FoveaHead and AtssHead.

* Update docs.

* Fix typos and docstrings.

* Fixes.

* Fix img_metas type in docstring.
2021-11-05 17:45:49 +08:00

2.4 KiB

OpenVINO Support

This tutorial is based on Linux systems like Ubuntu-18.04.

Installation

  1. Install OpenVINO.
  2. Install MMdeploy following the instructions.

Usage

Example:

python tools/deploy.py \
    configs/mmdet/single-stage/single-stage_openvino_dynamic.py \
    /mmdetection_dir/mmdetection/configs/ssd/ssd300_coco.py \
    /tmp/snapshots/ssd300_coco_20210803_015428-d231a06e.pth \
    tests/data/tiger.jpeg \
    --work-dir ../deploy_result \
    --device cpu \
    --log-level INFO \

List of supported models exportable to OpenVINO from MMDetection

The table below lists the models that are guaranteed to be exportable to OpenVINO from MMDetection.

Model name Config Dynamic Shape
FCOS configs/fcos/fcos_r50_caffe_fpn_gn-head_4x4_1x_coco.py Y
FSAF configs/fsaf/fsaf_r50_fpn_1x_coco.py Y
RetinaNet configs/retinanet/retinanet_r50_fpn_1x_coco.py Y
SSD configs/ssd/ssd300_coco.py Y
Faster R-CNN configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py Y
Cascade R-CNN configs/cascade_rcnn/cascade_rcnn_r50_fpn_1x_coco.py Y
Mask R-CNN configs/mask_rcnn/mask_rcnn_r50_fpn_1x_coco.py Y
Cascade Mask R-CNN configs/cascade_rcnn/cascade_mask_rcnn_r50_fpn_1x_coco.py Y
FoveaBox configs/foveabox/fovea_r50_fpn_4x4_1x_coco.py Y
ATSS configs/atss/atss_r50_fpn_1x_coco.py Y

Notes:

  • For faster work in OpenVINO in the Faster-RCNN, Mask-RCNN, Cascade-RCNN, Cascade-Mask-RCNN models the RoiAlign operation is replaced with the ExperimentalDetectronROIFeatureExtractor operation in the ONNX graph.