mmdetection

Commit Graph

Author	SHA1	Message	Date
Haian Huang(深度眸)	ff649bd870	add single model test scripts (#5375 ) * add single model test scripts * update * fix name error	3 years ago
BigDong	11f3ca2ba6	[Refactor] move model.pretrained to model.backbone.init_cfg (#5370 ) * [Refactor] move model pretrained to backbone init_cfg * fix some bug and add a test unit * add * fix bug * fix priority * fix bug * delete test backbone init cfg * fix potential bug * fix potential bug * add detectors test unit * fix logic * add test unit * fix	3 years ago
Maximilian Otten	eb13c97d2f	allow to configure out the mask_head of the HTC algorithm (#5389 ) The HTC algorithm has an interleaved execution of bbox_heads and mask_heads. For Object-Detection we only need the bbox_head. The design of the HTC algorithms allows to remove the mask_head while the bbox_head keeps working. Co-authored-by: Maximilian Otten <maximilian.otten@iml.fraunhofer.de>	3 years ago
Guangchen Lin	4c62867c9e	Delete deprecated warning in FPN (#5311 ) * deprecated warning in FPN * modify doc string * modify related files	3 years ago
Ye Liu	dd0e8ede1f	[Feature] Add support for mask diagonal flip in TTA (#5403 ) * Add support for mask diagonal flip in tta * Add unit test * Fix unit test * Fix unit test	3 years ago
沈云航 Yunhang Shen	102f379789	Some loss (Seesaw loss..) may have custom activation to get the predicted labels. (#5428 ) * Some loss (Seesaw loss..) may have custom activation to get the predicted labels. * change custom_cls_channels to custom_activation.	3 years ago
Shilong Zhang	e91da70114	Refactor anchor_generator and point_generator (#5349 ) * add sparse priors * add mlvlpointsgenerator * revert __init__ of core * refactor reppoints * delete label channal * add docstr * fix typo * fix args * fix typo * fix doc * fix stride_h * add offset * add offset * fix docstr * new interface of single_proir * fix device * add unitest * add cuda unitest * add more cuda unintest * fix reppoints * fix device * add unintest for ssd and yolo and rename prior_idxs * add docstr for MlvlPointGenerator * add space * add num_base_priors	3 years ago
Guangchen Lin	269bb9e8c6	fix default type (#5438 )	3 years ago
Haian Huang(深度眸)	0d80caf53e	fix link error of faster rcnn mstrain 3x (#5432 )	3 years ago
BigDong	0b2d8833d2	fix ssd_vgg bug (#5431 )	3 years ago
johnson-magic	d22e981132	fixbug (#5398 )	3 years ago
Junjun2016	e7fb275056	fix typos (#5415 )	3 years ago
v-qjqs	bbd03fb5d1	Supports for exporting DETR to onnx with dynamic shapes and batch inference (#5168 ) * support exporting `DETR` to ONNX with dynamic shapes and competitive performance * support exporting `DETR` to ONNX with dynamic shapes and competitive performance * add onnx performance docs for `DETR` * fix lint error * fix * refactor the onnx export for detr * fix doc * supports batch inference for detr * inherit config for batch inference * fix type * support batch inference for ONNX * fix dynamically clip bboxes * remove batch inference config	3 years ago
Guangchen Lin	206ce2ca71	[Docs] Modify pytorch2onnx docs (#5269 ) * modify onnx docs * modify onnx docs * fixgrammatical errors * we -> We * refer to -> following * use a more accurate reference * modify the expression * modify according to new get_start.md	3 years ago
Haian Huang(深度眸)	3072a122ca	[Enhancement]: add ImageNet Pretrained Models docs (#5268 ) * add ImageNet Pretrained Models docs * update * update md * update disc * update * Remove brackets	3 years ago
RangiLyu	4058255b46	[Enhancement]: Refactor SSD (#5291 ) * add vgg neck * refactor ssd neck and vgg * refactor ssd head and neck * init l2 norm * update config * change ssdvgg backbone * revert to SSD_VGG * add unit test * fix ssd voc * avoid BC-breaking * add TODO * add convert script * avoid BC breaking * update readme * update download link * Fix doc	3 years ago
jiangwenj02	d1ef85d9ff	Fix Grid-RCNN error when there is no bbox result (#5357 ) * fix * fix lint * fix flake8 * fix * fix Co-authored-by: MyCurveNet <137376864@qq.com>	3 years ago
Haian Huang(深度眸)	0e09f00a5b	[Enhance]: Update speed benchmark tools (#5265 ) * add calc fps file * update fps file * update test benchmark * update segm_mAP * update benchmark * update * update and add times per image outputs	3 years ago
RunningLeon	5ec72728a4	[Enhancement]: make deployment tools more friendly to use (#5280 ) * fisrt commit * convert with wrapper * update doc * remove std and mean from args * add deprecation warnings * add comments and deprecated warnings Co-authored-by: q.yao <yaoqian@sensetime.com>	3 years ago
Haian Huang(深度眸)	49b20ac420	fix gcnet x101 config error (#5360 ) * fix gcnet x101 config error * update gcnet dcn model link	3 years ago
Jiaqi Wang	acd023a890	Seesaw Loss - update bibtex (#5387 )	3 years ago
Dmytro Panchenko	87ad480998	Make YOLOv3 neck more flexible (#5218 ) * Make YOLOv3 neck more flexible * Fix linter warning * update docstring * add unit test of yolov3 neck * update * fix docstring Co-authored-by: RangiLyu <lyuchqi@gmail.com>	3 years ago
RangiLyu	48cfeb07a0	[Feature] Add Mask R-CNN 3x mstrain config (#5233 ) * add mstrain 3x models * fix * fix backbone * caffe norm * update cfg * update mask rcnn * update regnet mask rcnn * update model link * add ms3x coco instance cfg	3 years ago
RangiLyu	65882a3fc4	[Fix] Not use deprecated pretrained kwarg when it is None (#5363 ) * not use deprecated param when it is None * fix typo * fix typo	3 years ago
RangiLyu	a4e04e86d2	[Fix]Config path in metafile script (#5359 ) * fix mistakes * fix config path	3 years ago
Haian Huang(深度眸)	d5dbc0e2ef	[Feature] Add Faster R-CNN 3x mstrain config (#5179 ) * add faster_rcnn_r50_fpn_mstrain_3x_coco.py * add faster_rcnn_r50_fpn_mstrain_3x_coco_lr temp config * add img_scale * add RepeatDataset config * unified config * add faster rcnn config * refactor 3x config * update mstrain_3x_coco.py * add x101 * add x101_32_8d * fix x101_32_8d * add regnet mstrain 3x * fix lint * fix 3.2GF * fix caffe norm error * fix caffe norm error * Fix mean and std error in ResNeXt-101-32x8d * fix x101_32_4d * updata * fix unittest error	3 years ago
BigDong	991014edd7	[Docs] Supplement FAQ doc, add training loss=nan solution (#5312 ) * [Docs] Supplement FAQ doc, add training loss nan solution * [Fix] fix some error * Update faq.md Co-authored-by: Wenwei Zhang <40779233+ZwwWayne@users.noreply.github.com>	3 years ago
BigDong	c2e318dee0	[Docs] Add FAQ about COCO AP or AR =-1 (#5313 ) * [Docs] add faq about coco evaluation * [Docs] add faq about coco evaluation * [Docs] add FAQ about coco result=-1	3 years ago
Yezhen Cong	3affac5d55	Fix web links ending with `.md` (#5315 )	3 years ago
RangiLyu	969dbf4e0c	[Feature] Add dev_script to generate metafile. (#5325 ) * add pwc metafile script * support repeat dataset * generate download link * save in different yaml	3 years ago
Shilong Zhang	dc12b098e8	fix the import path of msda (#5338 ) * fix the import path of msda * fix the import path of msda	3 years ago
杨培文 (Yang Peiwen)	e0cb8666a5	change to https (#5328 )	3 years ago
Shilong Zhang	238d3f6edc	Revert the test_mixins to test image by image (#5249 ) * [Refactor] Add simple_test to dense heads * Move test methods of RPN * Fix assert * move fun and add docstr * Move parameter position * add yolact docstr * (1) Provide more detailed error information; (2) fix centernet aug error (3) fix point_rend aug error * change to single * add deprecated warning * revert bbox head * fix doc * add onnx_export to anchor_head * change the scale_factor to numpy * fix doc * remove postprocess of yolact * fix docs * fix docs * fix docstr * fix unitest * fix roi * fix #5211 * resolve comments Co-authored-by: Yosuke Shinya <42844407+shinya7y@users.noreply.github.com> Co-authored-by: hhaAndroid <1286304229@qq.com>	3 years ago
RangiLyu	0276d56ada	add install with mim (#5321 )	3 years ago
Wenwei Zhang	47d468ef08	Support pretrained models from MoCov2 (#5286 ) * support pretrained models from MoCov2 * support swav * update tag * add info * update model links * update	3 years ago
BigDong	3ea7905418	[Docs] clarify installation (#5316 )	3 years ago
Guangchen Lin	9def292c47	[Docs] Delete bug comment (#5277 ) * fix comment * update to the latest config format * modify comments	3 years ago
Wenwei Zhang	86dc7a3011	update pat container (#5303 )	3 years ago
Irtiza Hasan	0f3e8d7788	Updated the title for Pedestrian Detection: The Elephant In The Room and conference venue. (#5302 ) Updated title of Generalizable Pedestrian Detection: The Elephant In The Room --> Generalizable Pedestrian Detection: The Elephant In The Room. Conference venue to CVPR2021	3 years ago
Shilong Zhang	3126f95a97	Add simple_test to dense heads. (#5264 ) * [Refactor] Add simple_test to dense heads * Move test methods of RPN * Fix assert * move fun and add docstr * Move parameter position * add yolact docstr * (1) Provide more detailed error information; (2) fix centernet aug error (3) fix point_rend aug error * fix doc * remove postprocess of yolact * fix docs * fix docs Co-authored-by: Yosuke Shinya <42844407+shinya7y@users.noreply.github.com> Co-authored-by: hhaAndroid <1286304229@qq.com>	4 years ago
Yosuke Shinya	6ef9605bf3	Fix size mismatch in multiclass_nms (#4980 )	4 years ago
Shilong Zhang	a8c681d285	Bump version to v2.13.0 (#5262 ) * update changelog * update changelog * Improvements * Improvements * update for v2.13.0 Co-authored-by: hhaAndroid <1286304229@qq.com>	4 years ago
RunningLeon	8d40aefe04	[Feature]: Evaluation with TensorRT backend (#5198 ) * evaluate trt models * update version of onnx * update maskrcnn results * add backend argument * update fcos results * update * fix bug * update doc	4 years ago
杨培文 (Yang Peiwen)	52c935d27b	Fix torch.chunk int32 overflow (#5191 ) * Update fcn_mask_head.py * Update fcn_mask_head.py * Update fcn_mask_head.py * Resolve format issues Co-authored-by: Wenwei Zhang <40779233+ZwwWayne@users.noreply.github.com>	4 years ago
v-qjqs	2e635eb5f6	Supports for exporting CornerNet to ONNX with dynamic shapes and comparable performance (#5136 ) * Supports for exporting CornerNet to ONNX with dynamic shapes and comparable performance * add docs for exporting cornernet, and simplify code * fix doc * format doc * fix docstring	4 years ago
Shilong Zhang	24b6f9319f	revert coco_api to `6031ce1` and add comments (#5246 ) * fix get_cat_ids * pass tuple of str * change to not sort and add unitest * revert to `6031ce1a90`	4 years ago
Shilong Zhang	50b87bc312	fix `get_cat_ids` of coco_api (#5243 ) * fix get_cat_ids * pass tuple of str	4 years ago
Shilong Zhang	d30d0b688c	Make `cat_ids` be consistent with the order of `CLASSES` (#5227 ) * add commnets * add unitest * move to coco wapper	4 years ago
Shilong Zhang	6031ce1a90	refactor onnx export of two stage (#5205 ) * refactor rpn * refactor bbox_head * mask export * add docstr for onnx export * support onnx * fix several bug related to onnx * fix onnx * fix onestage * fix dicstr * add more doc for bbox head * fix unitest * fix scale factor of two stage * fix type * change to np.array * fix a bug of maskrcnn * fix typeg * resolve comments * fix doc * fix unitest * resolve conversation * fix docstr * resolve conversation * fix docstr * fix docstr	4 years ago
RangiLyu	d3ea2f3d0a	Update YOLOv3 models (#5229 ) * update yolov3 models * update model links	4 years ago

1 2 3 4 5 ...

664 Commits (ff649bd870190567d6c60aed4286909d5370da7b) All Branches Search

664 Commits (ff649bd870190567d6c60aed4286909d5370da7b)

All Branches