Iteration Plan v1.3.15 - Oct 2021

[zhouzaida]

MMCV v1.3.14 has been released, see the release notes
We list our plan for v1.3.15, which will be released on October 15th.
The PRs from the communities will also be listed here.
Thus, developers could comprehensively know the plan and the progress.

Features:

• [Feature] Loading objects from different backends and dumping objects to different backends #1330
• [Feature] Upload checkpoints and logs to ceph #1375
• [Feature] Add fps op from mmdet3d #1337
• [Feature] Add gather points op from mmdet3d #1338
• [Feature] Add knn op from mmdet3d #1354
• [Feature] Add interpolate ops from mmdet3d #1355
• [Feature] Add iou3d op from mmdet3d #1356
• [Feature] Add assign_score_withk op for paconv from mmdet3d #1357
• [Feature] Add roipoint_pool3d op from mmdet3d #1358
• Add Deformable Conv CustomOp for onnxruntime #1343
• [Feature] Add voxel ops from mmdet3d #1381
• [Feature] Add roiaware pool3d ops from mmdet3d #1382
• [Feature] Add spconv ops from mmdet3d #1385

Enhancements:

• [Enhancement] Raising warning when the different number of GPU is set for resuming #1360
• [Refactor] Remove some outdated lines #1370

Documentations:

• [Docs] cnn.md translated #1246
• [Docs] Add zh_cn document of ONNX #1331
• [Docs] Add docs_zh_CN of TensorRT #1336
• [Docs] Translate building documentation #1353
• [Docs] Translate contributing documentation #1351
• Polish readme #1320

Fixes:

• fix MultiScaleDeformableAttention inference issue on cpu model #1223
• [Fix] Revise unit test of correlation #1368
• [Fix] Add load_url to handle incompatibility of PyTorch versions #1377
• [Fix] Fix the weight initialization of deform conv #1262
• [Enhancement] Add error handler when getting host failed #1374
• Fix DCN and Modulated DCN error in parrots #1378

Others:

• Switch to the docker image provided by PyTorch #1376
• Add code spell hook to pre-commit and fix typos #1384

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.