Update saliency code

.devcontainer/devcontainer.json

@@ -1,11 +1,7 @@
 // See this page for reference of options: https://containers.dev/implementors/json_reference
 {
  "name": "Existing Dockerfile",
-<<<<<<< Updated upstream
  "image": "ghcr.io/tritonuas/obcpp:x86",
-=======
- "image": "ghcr.io/tritonuas/obcpp:jetson",
->>>>>>> Stashed changes
  // enable when need to connect over USB to pixhawk
  // also: need to run obcpp with sudo or add tuas user to dialout group with
  // `sudo usermod -aG dialout tuas && newgrp && bash`

docker/Dockerfile.jetson

@@ -1,12 +1,4 @@
-<<<<<<< Updated upstream
 FROM dustynv/l4t-pytorch:r36.2.0
-=======
-# FROM tritonuas/jetson-base:r36.2.0
-FROM dustynv/l4t-pytorch:r36.2.0
-
-# this base image came from dusty-nv/jetson-containers
-# ./build.sh --name=tritonuas/jetson-base pytorch:2.1 torchvision opencv 
->>>>>>> Stashed changes
 
 ARG USERNAME=tuas USER_UID=1000 USER_GID=1000 DEBIAN_FRONTEND=noninteractive
 
@@ -58,14 +50,6 @@ RUN --mount=target=/var/lib/apt/lists,type=cache,sharing=locked \
  libopenblas-dev \
  ninja-build
 
-# RUN sudo dpkg --remove libopencv-dev
-# RUN sudo apt install -f
-# RUN --mount=target=/var/lib/apt/lists,type=cache,sharing=locked \
-# --mount=target=/var/cache/apt,type=cache,sharing=locked \
-# rm -f /etc/apt/apt.conf.d/docker-clean \ 
-# && apt-get update \
-# && apt-get install -y libopencv-dev
-
 RUN pip3 install typing-extensions PyYAML cpplint
 
 RUN echo $USERNAME ALL=\(root\) NOPASSWD:ALL > /etc/sudoers.d/$USERNAME \
@@ -76,12 +60,11 @@ RUN git clone --depth 1 https://github.com/mavlink/MAVSDK.git --branch v2.9.1 --
  && cd MAVSDK \
  && git submodule update --init --recursive \
  && cmake -DCMAKE_BUILD_TYPE=Release -Bbuild/default -H. \
- && cmake --build build/default -j2 --target install
+ && cmake --build build/default -j`nproc` --target install
 
 # pull and build torchvision
 # refer to this page for version compatibilty with pytorch (libtorch) https://github.com/pytorch/pytorch/wiki/PyTorch-Versions
 ARG TORCHVISION_VERSION=0.17.0
-<<<<<<< Updated upstream
 # Space separated list of CUDA architecture versions.
 # The version nubmers depend on the NVIDIA GPU model we're using and the installed CUDA version. 
 # For the Jetson Orin Nano with the "Ampere" architecture and CUDA 12.1 we can use version 8.6 (written as 86 in CUDA_ARCH_LIST).
@@ -97,7 +80,7 @@ RUN wget "https://github.com/pytorch/vision/archive/refs/tags/v${TORCHVISION_VER
  && mkdir build \
  && cd build \
  && cmake -DWITH_CUDA=1 -DTORCH_CUDA_ARCH_LIST="${CUDA_ARCH_LIST}" -DCUDA_HAS_FP16=1 -DCUDA_NO_HALF_OPERATORS=1 -DCUDA_NO_HALF_CONVERSIONS=1 -DCUDA_NO_HALF2_OPERATORS=1 -DCMAKE_BUILD_TYPE=Release -DCMAKE_PREFIX_PATH="/usr/local/lib/python3.10/dist-packages/torch/share/cmake/Torch" .. \
- && make -j4 \
+ && make -j`nproc` \
  && make install
 
 RUN pip3 install gdown
@@ -111,45 +94,18 @@ RUN gdown 1VtBji-cWfetM5nXZwt55JuHPWPGahQOH -O ${ARENA_TAR_PATH}
 RUN tar -xvzf ${ARENA_TAR_PATH}
 WORKDIR ${ARENA_EXTRACTED_PATH}
 RUN sh Arena_SDK_ARM64.conf
-=======
-#ARG TORCHVISION_INSTALL_DIR=/torchvision-tmp
-WORKDIR ${TORCHVISION_INSTALL_DIR} 
-RUN wget "https://github.com/pytorch/vision/archive/refs/tags/v${TORCHVISION_VERSION}.zip" \
- && unzip "v${TORCHVISION_VERSION}.zip" \
- && cd vision-0.17.0 \
- && mkdir build \
- && cd build \
- && cmake -DWITH_CUDA=1 -DCUDA_HAS_FP16=1 -DCUDA_NO_HALF_OPERATORS=1 -DCUDA_NO_HALF_CONVERSIONS=1 -DCUDA_NO_HALF2_OPERATORS=1 -DCMAKE_BUILD_TYPE=Release -DCMAKE_PREFIX_PATH="/usr/local/lib/python3.10/dist-packages/torch/share/cmake/Torch" .. \
- && make -j4 \
- && make install
-
-# # Install g++10 and replace the older version. For some reason some c++ 20 features aren't working with g++9 even though
-# # we have CMake configured to use c++ 20 https://stackoverflow.com/questions/6903.1073/why-am-i-missing-c20-headers-and-how-do-i-fix-this
-# RUN apt-get update && apt-get install -y g++-10 gcc-10
-# RUN update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-10 10
-# RUN update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-10 10
-# RUN update-alternatives --set gcc /usr/bin/gcc-10
-# RUN update-alternatives --set g++ /usr/bin/g++-10
->>>>>>> Stashed changes
 
 WORKDIR /obcpp
 COPY . .
 
 RUN rm -rf /obcpp/build
 WORKDIR /obcpp/build
 ENV CMAKE_PREFIX_PATH="/usr/local/lib/python3.10/dist-packages/torch/share/cmake/Torch;/usr/local/share/cmake/TorchVision"
-<<<<<<< Updated upstream
 RUN GITHUB_ACTIONS=true cmake -DCMAKE_PREFIX_PATH="/usr/local/lib/python3.10/dist-packages/torch/share/cmake/Torch;/usr/local/share/cmake/TorchVision" -DCMAKE_MODULE_PATH="/usr/local/share/cmake/TorchVision" -DCMAKE_BUILD_TYPE="Release" ..
 
 RUN ninja obcpp 
-=======
-RUN GITHUB_ACTIONS=true cmake -DCMAKE_PREFIX_PATH="/usr/local/lib/python3.10/dist-packages/torch/share/cmake/Torch;/usr/local/share/cmake/TorchVision" -DCMAKE_MODULE_PATH="/usr/local/share/cmake/TorchVision" -DCMAKE_BUILD_TYPE="Release" -DCMAKE_JOB_POOLS="j=2" ..
-
-# RUN make obcpp cuda_check load_torchvision_model VERBOSE=1
-RUN ninja obcpp
->>>>>>> Stashed changes
 
 # login as non-root user
 # USER $USERNAME
 
-CMD [ "/obcpp/build/bin/obcpp" ]
+CMD [ "/obcpp/build/bin/obcpp", "/obcpp/configs/config.json"] 

docker/Makefile

@@ -2,20 +2,23 @@
 all: 
  echo "select one of the other targets"
 
+build-x86-image:
+ docker buildx build --platform linux/amd64 --tag ghcr.io/tritonuas/obcpp:x86 --file Dockerfile.x86 ..
+
 # NOTE: the jetson Docker image is based on a custom base image we generate from 
 # https://github.com/dusty-nv/jetson-containers/
 # - Clone that repo and run the following command: 
 # ./build.sh --name=tritonuas/jetson-base pytorch:2.1 torchvision opencv 
 # - Push that image to dockerhub as whatever tag the FROM line in our Dockerfile.jetson is
 # - Run the following build-jeston-image target
 build-jetson-image:
- docker buildx build --platform linux/arm64 --tag tritonuas/obcpp:jetson --file Dockerfile.jetson ..
-
-push-jetson-image:
- docker push tritonuas/obcpp:jetson
+ docker buildx build --platform linux/arm64 --tag ghcr.io/tritonuas/obcpp:jetson --file Dockerfile.jetson ..
 
 build-arm-image:
- DOCKER_BUILDKIT=1 docker build --tag tritonuas/obcpp:arm --file Dockerfile.arm ..
+ docker buildx build --platform linux/arm64 --tag ghcr.io/tritonuas/obcpp:arm --file Dockerfile.arm ..
+
+push-jetson-image:
+ docker push ghcr.io/tritonuas/obcpp:jetson
 
 run-jetson-pixhawk-compose:
  docker compose -f jetson-pixhawk-compose.yml up -d
@@ -24,7 +27,14 @@ stop-jetson-pixhawk-compose:
  docker compose -f jetson-pixhawk-compose.yml down
 
 run-jetson-cuda-check:
- docker run -it --rm --runtime nvidia -i tritonuas/obcpp:nvidia /obcpp/build/bin/cuda_check
+ docker run -it --rm --runtime nvidia -i ghcr.io/tritonuas/obcpp:jetson /obcpp/build/bin/cuda_check
+
 
+# NOTE: Use this target for development where you want to quicly edit source code and recompile.
+# This will spawn up the jetson container, launch you into a bash shell and mount the 
+# host's obcpp directory at "/obcpp" in the container. This means you can edit the source 
+# files on the host (with VSCode over SSH or Vim) and recompile in the Docker environment.
+# Note that to re-run cmake inside the container you'll need the really long CMake command
+# in the Dockerfile.jetson.
 jetson-develop:
- cd .. && docker run -it --net=host --runtime=nvidia --volume=./:/obcpp -i tritonuas/obcpp:jetson /bin/bash
+ cd .. && docker run -it --net=host --runtime=nvidia --platform=linux/arm64 --volume=./:/obcpp --device=/dev/ttyACM0 tritonuas/obcpp:jetson /bin/bash

include/cv/saliency.hpp

@@ -36,7 +36,6 @@ class Saliency {
  private:
  std::string modelPath; // path to prediction model
  torch::jit::script::Module module; // the loaded model
- // c10::Device device; // 
 };
 
 #endif // INCLUDE_CV_SALIENCY_HPP_

src/cv/saliency.cpp

@@ -26,7 +26,6 @@ Saliency::Saliency(std::string modelPath) {
  catch (const c10::Error& e) {
  LOG_F(ERROR, "error loading the model %s", e.msg().c_str());
  }
-
 }
 
 std::vector<CroppedTarget> Saliency::salience(cv::Mat image) {
@@ -37,7 +36,8 @@ std::vector<CroppedTarget> Saliency::salience(cv::Mat image) {
  // swap axis
  tensor = Saliency::transpose(tensor, { (2), (0), (1) });
 
- c10::Device device = torch::cuda::is_available() ? torch::kCUDA : torch::kCPU; // eventually add device as member of Saliency
+ // eventually add device as member of Saliency
+ c10::Device device = torch::cuda::is_available() ? torch::kCUDA : torch::kCPU;
  auto tensor_cuda = tensor.to(device);
 
  auto input_to_net = ToInput(tensor_cuda);
@@ -48,7 +48,7 @@ std::vector<CroppedTarget> Saliency::salience(cv::Mat image) {
  * that we want are : a) boxes (FloatTensor[N, 4]): the predicted boxes, and
  * b) scores (Tensor[N]): the scores of each detection.
  */
- 
+
  // output is a tuple of (losses, detections)
  auto output = module.forward(input_to_net);
  c10::ivalue::Tuple& tuple = output.toTupleRef();