Iteration Plan of v2.17.0 - September 2021

[RangiLyu]

Iteration Plan of v2.17.0 - September 2021

MMDetection v2.16.0 has been released here.
We list our plan for v2.17.0, which will be released on September 27th.
The PRs from the communities will also be listed here.
Thus, developers could comprehensively know the plan and the progress.

Algorithms:

• [Feature]SOLO: Segmenting Objects by Locations #5832
• [Feature] Add pvt and pvtv2 #5780
• [Feature]: Add swin coco pretrained models. #6100
• [Feature] Support QueryInst #6050

Refactors:

• Refactor export One-Stage model to ONNX #5251

Features:

• [Enhancement]: Exporting some models to OpenVINO #5924
• [Feature]Add base one-stage segmentor #5904
• [Feature]Add a general data structrue for the results of models #5508

Enhancements:

• [Enhance]: Check dtype in transform unit tests. #5969
• YOLOX supports fp16 training #5983
• YOLACT supports val workflow #5986
• [Enhance]: Speedup yolov3 inference. #5991

Documentation:

• [Docs]Add Chinese translation of docs_zh-CN/tutorials/customize_dataset.md #5915
• [Enhance] Add dataset information in the README file for PanopticFPN #5995

Fixes:

• Fix empty proposal errors in the training of some two-stage models #5941
• Fix regression loss does not support empty tensor problem. #5976
• fix dynamic shape bug #5925
• [Enchance] Add filter rules of Mosaic transform #5897

Community help wanted:

• Dataset download scripts. (是否考虑增加脚本 #4932)
• CrowdDet implementation. (https://github.com/Purkialo/CrowdDet)

If you have created a PR and wish it to be merged in this version, feel free to leave a message in this issue. We will update our plan accordingly here.