a2f1bcb68e
[cherry-pick] Support cmake configure when system exists multiple cuda versions. ( #1760 )
...
* update cmake
* typos
2023-02-14 18:24:10 +08:00
44b144f095
[Cherry-pick] Fix `WarpBbox` and memory leak in `TextRecognizer` ( #1752 )
...
(cherry picked from commit cadc2658f3
2023-02-13 19:55:36 +08:00
f7fdcb4aa0
fix instance norm double free ( #1749 )
2023-02-13 19:42:54 +08:00
2efb42d8d8
[Enhancement] check in tensorrt-fp16 deployment config for simcc ( #1741 )
...
* check in tensorrt-fp16 deployment config for simcc
* update
2023-02-09 20:11:53 +08:00
19815ae9ae
add coco whole-body skeleton ( #1742 )
...
(cherry picked from commit 31b099a37b
2023-02-09 19:10:07 +08:00
8e254bc846
enable TRT parse ONNX model from file ( #1734 )
2023-02-09 15:55:04 +08:00
a8f8c4febe
[Refactor] Support batch inference with shape clustering ( #1733 )
...
* refactor `NetModule`
* name
* fix sorting
* fix indices
(cherry picked from commit f5a05b5219
2023-02-08 20:36:29 +08:00
bc1b6440cd
fix mmocr create input error when the arg is array ( #1721 )
2023-02-08 13:47:47 +08:00
d28020ec34
[Docs] Update tutorial "introduction to model deployment" ( #1709 )
...
* update docs
* update
2023-02-08 11:36:06 +08:00
c3b6c37464
Cherry pick PR #1625 , #1715 and #1725
...
Cherry-pick PR 1625, 1715, 1725 to dev-1.x
2023-02-08 11:28:05 +08:00
2731b61abd
[Fix] Add an option to flip webcam inputs for pose tracker demo ( #1725 )
...
(cherry picked from commit 5de0ecfcaf
2023-02-07 21:12:44 +08:00
682cb79bc5
[Enhancement] Optimize C++ demos ( #1715 )
...
* optimize demos
* show text in image
* optimize demos
* fix minor
* fix minor
* fix minor
* install utils & fix demo file extensions
* rename
* parse empty flags
* antialias
* handle video complications
(cherry picked from commit 2b18596795
2023-02-07 21:12:44 +08:00
1f56eea807
[Feature] Support feature map output for mmsegmentation ( #1625 )
...
* add feature map output for mmseg
* update api
* update demo
* fix return
* update format_shape
* fix lint
* update csharp demo
* update python demo && api
* fix coreml build
* fix lint
* better sort
* update
* update cpp demo & add missing header
* change to CHW
* update csharp api
* update isort version to 5.12.0
* fix python api
* fix log
* more detail api docs
* isort support python3.7
* remove isort change
* remove whitespace
* axes check
* remove FormatShapeImpl
* minor
* add permute tc
* remove stride buffer
(cherry picked from commit b85f34141b
2023-02-07 21:04:20 +08:00
a9f8d8c951
fix torch allocator resouce releasing ( #1711 )
2023-02-06 11:57:56 +08:00
87e48140f4
[Feature] Support RTMDet-Ins. ( #1662 )
...
* [Feature] Support RTMDet-Ins.
* fix visualize
* fix rewrite trt
* add config
* support torch 1.13
* fix keep ratio resize
* resolve scale factor bug
* set to None
2023-02-06 11:06:27 +08:00
3e6f2b3f47
fix error: build_backend_model() takes from 1 to 2 positional arguments but 3 were given ( #1705 )
2023-02-03 13:31:13 +08:00
8697dacd8d
[Fix] fix mmaction2 visualizer ( #1703 )
...
* fix mmaction2 visualizer
* fix lint
2023-02-03 10:20:20 +08:00
db9e633b68
[Doc] Correct cloned branch in build document ( #1700 )
2023-02-03 10:19:43 +08:00
a62ed41f44
[Fix] miss importing torchvision in torchscript wrapper ( #1702 )
2023-02-02 14:11:16 +08:00
153682e804
[Fix] add bounds to avoid large resource usage of nms operator on jetson ( #1687 )
...
* fix nms trt jetson
* fix comment
* clang format
2023-02-01 15:45:22 +08:00
c3fda1905b
[Enhancement] Add mmyolo CI on mmdeploy ( #1650 )
...
* add mmyolo ut workflow
* add test_linux/windows ort/trt for mmyolo
* fix abs path to rel path
* fix mmyolo ut
* fix mmyolo position
* fix mmyolo dir
* fix build yolo
* fix yolo path
* fix yolo path
* fix yolo install path
* fix yolo install path for cuda
* remove dumplicated ci
* reformat script
2023-02-01 14:49:59 +08:00
5ab0bcb5ea
[Fix] fix rknn quantization pipeline ( #1636 )
...
* fix rknn quantization pipeline
* resolve comments
2023-01-31 16:35:11 +08:00
a19cf64765
Fix mapping error from mmpose's regression codec to SDK's components ( #1684 )
2023-01-31 15:58:04 +08:00
2ceabd0274
[Enhancement] Add 320x320 static config for tensorrt ( #1689 )
...
* add trt static 320 config for rtmdet
* only add on detection folder
2023-01-31 14:36:17 +08:00
8062e4e41d
[Enhancement] remove MMDEPLOY_BUILD_SDK_CXX_API option ( #1688 ) ( #1692 )
...
* remove MMDEPLOY_BUILD_SDK_CXX_API option
* update
* update
(cherry picked from commit b101a4af65
2023-01-31 14:25:22 +08:00
3bf1336262
Cherry pick PR #1663 , #1666 , #1486 to dev-1.x
...
Cherry pick PR 1663, 1666, 1486 to dev-1.x
2023-01-31 13:59:00 +08:00
dc1d9df315
[Fix] fix coreml for branch 1.x ( #1669 )
...
* fix coreml for branch 1.x
* fix docstring
* update docsting1
2023-01-31 13:49:48 +08:00
0c1adba988
Fix isort lint error by upgrading it to 5.11.5 ( #1680 )
2023-01-31 13:32:41 +08:00
7d96969149
[Feature] Pose tracker C/C++/Python API&demos ( #1663 )
...
* add PoseTracker API
* add mahalanobis distance, add det_pose demo
* simplify api
* simplify api
* fix cmake & fix `CropResizePad`
* ignore out of frame bboxes
* clean-up
* fix lint
* add c api docs
* add c++ api docs/comments
* fix gcc7 build
* fix gcc7+opencv3
* fix stupid lint
* fix ci
* add help info & webcam support for C++ pose tracker demo
* add webcam support for Python pose tracker demo
* fix lint
* minor
* minor
* fix MSVC build
* fix python binding
* simplify module adapter
* fix module adapter
* minor fix
(cherry picked from commit 3d425bbb9f
2023-01-31 12:07:40 +08:00
d91ba2f3c8
fix 'cvtcolor' error in the preprocessing of single channel images ( #1666 )
...
(cherry picked from commit 8bb3fcc6d8
2023-01-31 12:07:40 +08:00
8d344ee164
[Enhancement] Speedup TopDownAffine by CropResizePad ( #1486 )
...
* "use 'CropResizePad' to speed up topdownaffine"
* add missing header
(cherry picked from commit c458e2a524
2023-01-31 12:07:40 +08:00
878a364470
fix rknn output index error in SDK ( #1682 )
2023-01-30 20:51:06 +08:00
cbddf5a458
[Fix] Fix preprocess_model_config for CIFAR dataset ( #1659 )
...
* fix cifar10 for mmcls
* remove unnecessary code
2023-01-18 16:31:11 +08:00
15ad234a7a
Cherry-pick PR 1366, 1370, 1460, 1577, 1598, 1601, 1612, 1626 to dev-1.x ( #1635 )
...
* [Fix] Fix example standalone build for msvc (#1626 )
* fix example build for msvc
* move /Zc:__cplusplus to core
* [FIX] Fix csharp net48 and batch inference (#1370 )
* fix csharp net48
* add missing file
* update
* fix batch inference
* update demo
* update
* update version
* update nuget package version & fix project name
* build monolithic SDK by default (#1577 )
* Fix debug build for PoseTracker (#1601 )
* [Fix] Fix aligned allocations on Android (#1612 )
* fix android alignment
* fix typo
* fix size
* Fix build error on windows-cuda platform (#1598 )
* [Fix] Relax module adapter template constraints (#1366 )
* relax module adapter constraint
* remove forwarding `operator()`
* fix YOLOV5Head naming
Co-authored-by: Li Zhang <lzhang329@gmail.com>
2023-01-17 18:52:58 +08:00
953d33cc09
[Fix] fix headless device visualize ( #1630 )
...
* fix headless device visualize
* remove pretrain
* sync with 1641
* better show
2023-01-17 11:00:24 +08:00
ff95bf48ea
[Feature] Support more MMOCR models: DBNet++, TextSnake, ABINet, MRCNN ( #1534 )
...
* WIP ocr
* add mrcnn rewrite
* add any rewrite for abinet
* export abinet to onnx
* fix abinet onnx export
* support abinet to tensorrt static and modify mmocr.yml
* add textsnake and dbnetpp
* support mrcnn in ORT and TRT
* add a condition before update data_preprocessor scope
* update doc and mmocr.yml
* add ut
* markdown and simple config
* write build_pytorch_model in child class
* update any_default
* remove where in abi_language_decoder___get_length__default
* keep where
* fix UT
* fix UT
* fix UT
* update mmocr.yml and config description
* tensorrt-fp32 -> tensorrt
* update doc
2023-01-16 12:06:03 +08:00
7e66cfc59b
Cherry-pick PR #1460 : optimize pose tracker
...
Cherry-pick PR 1460 to dev-1.x
2023-01-16 10:57:06 +08:00
ef260d8df5
[Fix] Substitute SyncBN with BN for mmpose ( #1631 )
...
* fix syncbn for mmpose
* unify build_pytorch_model
2023-01-16 10:50:45 +08:00
89580eb496
cherry-pick #1608 : change log file extension to '.txt' in regression test( #1637 )
2023-01-13 16:12:37 +08:00
11e8a2179e
Fix the performance test for RKNN ( #1583 )
2023-01-11 11:30:04 +08:00
6b9bfcfceb
[Fix]: fix numba numpy version not compatiable in circileci ( #1611 )
...
* fix numba numpy version not compatibale
* fix mmcls ut
* update
* update
* only do lint when deploy configs changed
2023-01-11 10:48:52 +08:00
92efd9cb7b
[Doc] update build doc for jetsons ( #1621 )
...
* update build doc for jetsons
* set lint only true for docs
* change base revision of circleci config
Co-authored-by: RunningLeon <mnsheng@yeah.net>
2023-01-10 20:03:32 +08:00
3e22f96f04
Fixed calling miniconda script in Dockerfile ( #1629 )
2023-01-10 15:50:15 +08:00
0cfee8c0f4
fix no test_step error in TorchWrapper ( #1600 )
2023-01-10 10:53:38 +08:00
f7ea130a6f
fix android build command
2023-01-06 16:11:56 +08:00
71fc8e3241
CodeCamp #101 : Support MMDetection 3.x RTMDet model deployment on RV1126 ( #1551 )
...
* * partition rtmdet
* * add rtmdet deploy config
* * add rtmdet deploy config
* * modify rtmdet pipline anchor_generator's info dump
* support rtmdet infer in sdk
* fix a bug
* * fix a bug in csrc/mmdeploy/preprocess/transform/normalize.cpp
* * fix a bug
* * update docs
* * fix lint
* * update several urls in docs
2023-01-06 14:01:31 +08:00
db8de7ec1f
[Fix] Fix typo in 'LetterResize' transform ( #1618 )
2023-01-06 11:24:54 +08:00
5d0a08eafb
fix log info typo of check env ( #1613 )
2023-01-06 11:10:15 +08:00
f288d00d50
[Enhancement] Optimize pose tracker ( #1460 )
...
* sync master
* suppress overlapped tracks
* add CUDA WarpAffine
* export symbols
* fix linkage
* update pose tracker
* clean-up
* fix MSVC build
* fix MSVC build
* add ffmpeg cli command
2023-01-05 11:57:21 +08:00
14550aac38
[Enhancement] Switch pip to mim in Docs and Dockerfile ( #1591 )
...
* change pip install to mim install in docs and Dockerfil
* merge with latest dev-1.x
2023-01-04 23:11:13 +08:00
Chen Xin
Li Zhang
q.yao
lvhan028
AllentDan
Xiang Xu
Chen Xin
RangiLyu
wanglin
hanrui1sensetime
tripleMu
RunningLeon
Nghia
hanrui1sensetime
Qingren
Xin Li