mirror of
https://github.com/open-mmlab/mmdeploy.git
synced 2025-01-14 08:09:43 +08:00
fd40119694
* fix custom ops support, fix multiple mark bug, add name mapping * check if the value_info need to be added * remove unnecessary print * add nms implement * two stage split wip * add two stage split * add split retinanet visualize * add two stage split (wip) * finish two stage split * fix lint * move parse string to mmdeploy.utils * add func mark count dict * use assert_cfg_valid * update func count before add Mark * fix dynamic shape support * add calib data generator * create calib dataset * finish end2end int8 * add split two stage tensorrt visualize
MMDeployment
Installation
-
Build backend ops
-
update submodule
git submodule update --init -
Build with onnxruntime support
mkdir build cd build cmake -DBUILD_ONNXRUNTIME_OPS=ON -DONNXRUNTIME_DIR=${PATH_TO_ONNXRUNTIME} .. make -j10 -
Build with tensorrt support
mkdir build cd build cmake -DBUILD_TENSORRT_OPS=ON -DTENSORRT_DIR=${PATH_TO_TENSORRT} .. make -j10 -
Build with ncnn support
mkdir build cd build cmake -DBUILD_NCNN_OPS=ON -DNCNN_DIR=${PATH_TO_NCNN} .. make -j10 -
Or you can add multiple flags to build multiple backend ops.
-
-
Setup project
python setup.py develop
Usage
python ./tools/deploy.py \
${DEPLOY_CFG_PATH} \
${MODEL_CFG_PATH} \
${MODEL_CHECKPOINT_PATH} \
${INPUT_IMG} \
--work-dir ${WORK_DIR} \
--device ${DEVICE} \
--log-level INFO