chris/frigate
1
0
Fork 0
forked from Github/frigate
Code Pull Requests Packages Activity

Upgrade TensorRT to 8.5.3 (#7006)

Browse Source 
* Update to latest tensorrt (8.6.1) release

* Build trt libyolo_layer.so in container

* Update tensorrt_models script to convert models from the frigate container

* Fix typo in model script

* Fix paths to yolo lib and models folder

* Add S6 scripts to test and convert specified TensortRT models at startup.

Rearrange tensorrt files into a docker support folder.

* Update TensorRT documentation to reflect the new model conversion process and minimum HW support.

* Fix model_cache path to live in config directory

* Move tensorrt s6 files to the correct directory

* Fix issues in model generation script

* Disable global timeout for s6 services

* Add version folder to tensorrt model_cache path

* Include TensorRT version 8.5.3

* Add numpy requirement prior to removal of np.bool

* This TRT version uses a mixture of cuda dependencies

* Redirect stdout from noisy model conversion
This commit is contained in:
Nate Meyer
2023-07-06 15:20:33 -04:00
committed by GitHub
parent 30dfdf47d4
commit dd02958f7c
14 changed files with 125 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
Dockerfile
View File

@@ -262,15 +262,35 @@ FROM deps AS frigate
WORKDIR /opt/frigate/
COPY --from=rootfs / /
# Build TensorRT-specific library
FROM nvcr.io/nvidia/tensorrt:23.03-py3 AS trt-deps
RUN --mount=type=bind,source=docker/support/tensorrt_detector/tensorrt_libyolo.sh,target=/tensorrt_libyolo.sh \
/tensorrt_libyolo.sh
# Frigate w/ TensorRT Support as separate image
FROM frigate AS frigate-tensorrt
#Disable S6 Global timeout
ENV S6_CMD_WAIT_FOR_SERVICES_MAXTIME=0
ENV TRT_VER=8.5.3
ENV YOLO_MODELS="yolov7-tiny-416"
COPY --from=trt-deps /usr/local/lib/libyolo_layer.so /usr/local/lib/libyolo_layer.so
COPY --from=trt-deps /usr/local/src/tensorrt_demos /usr/local/src/tensorrt_demos
COPY docker/support/tensorrt_detector/rootfs/ /
RUN --mount=type=bind,from=trt-wheels,source=/trt-wheels,target=/deps/trt-wheels \
pip3 install -U /deps/trt-wheels/*.whl && \
ln -s libnvrtc.so.11.2 /usr/local/lib/python3.9/dist-packages/nvidia/cuda_nvrtc/lib/libnvrtc.so && \
ldconfig
# Dev Container w/ TRT
FROM devcontainer AS devcontainer-trt
COPY --from=trt-deps /usr/local/lib/libyolo_layer.so /usr/local/lib/libyolo_layer.so
COPY --from=trt-deps /usr/local/src/tensorrt_demos /usr/local/src/tensorrt_demos
COPY docker/support/tensorrt_detector/rootfs/ /
COPY --from=trt-deps /usr/local/lib/libyolo_layer.so /usr/local/lib/libyolo_layer.so
RUN --mount=type=bind,from=trt-wheels,source=/trt-wheels,target=/deps/trt-wheels \
pip3 install -U /deps/trt-wheels/*.whl