trainingsample 0.2.16__tar.gz → 0.2.18__tar.gz

{trainingsample-0.2.16 → trainingsample-0.2.18}/.github/workflows/ci.yml

@@ -32,12 +32,12 @@ jobs:
         ls -la /usr/lib/llvm-18/lib/libclang.so
         which llvm-config-18
 
-        # Build OpenCV 4.12 from source since Ubuntu 24.04 only has 4.6
-        echo "Building OpenCV 4.12 from source..."
+        # Build OpenCV 4.11 from source since Ubuntu 24.04 only has 4.6
+        echo "Building OpenCV 4.11 from source..."
         cd /tmp
-        wget -q https://github.com/opencv/opencv/archive/4.12.0.tar.gz
-        tar -xzf 4.12.0.tar.gz
-        cd opencv-4.12.0
+        wget -q https://github.com/opencv/opencv/archive/4.11.0.tar.gz
+        tar -xzf 4.11.0.tar.gz
+        cd opencv-4.11.0
         mkdir build
         cd build
         cmake \
@@ -127,12 +127,12 @@ jobs:
         ls -la /usr/lib/llvm-18/lib/libclang.so
         which llvm-config-18
 
-        # Build OpenCV 4.12 from source since Ubuntu 24.04 only has 4.6
-        echo "Building OpenCV 4.12 from source..."
+        # Build OpenCV 4.11 from source since Ubuntu 24.04 only has 4.6
+        echo "Building OpenCV 4.11 from source..."
         cd /tmp
-        wget -q https://github.com/opencv/opencv/archive/4.12.0.tar.gz
-        tar -xzf 4.12.0.tar.gz
-        cd opencv-4.12.0
+        wget -q https://github.com/opencv/opencv/archive/4.11.0.tar.gz
+        tar -xzf 4.11.0.tar.gz
+        cd opencv-4.11.0
         mkdir build
         cd build
         cmake \
@@ -218,12 +218,12 @@ jobs:
         ls -la /usr/lib/llvm-18/lib/libclang.so
         which llvm-config-18
 
-        # Build OpenCV 4.12 from source since Ubuntu 24.04 only has 4.6
-        echo "Building OpenCV 4.12 from source..."
+        # Build OpenCV 4.11 from source since Ubuntu 24.04 only has 4.6
+        echo "Building OpenCV 4.11 from source..."
         cd /tmp
-        wget -q https://github.com/opencv/opencv/archive/4.12.0.tar.gz
-        tar -xzf 4.12.0.tar.gz
-        cd opencv-4.12.0
+        wget -q https://github.com/opencv/opencv/archive/4.11.0.tar.gz
+        tar -xzf 4.11.0.tar.gz
+        cd opencv-4.11.0
         mkdir build
         cd build
         cmake \
@@ -265,7 +265,7 @@ jobs:
         components: rustfmt, clippy
 
     - name: Install cargo-audit
-      run: cargo install cargo-audit
+      run: cargo install cargo-audit --version 0.22.2 --locked
 
     - name: Install pre-commit
       run: pip install pre-commit
@@ -298,12 +298,12 @@ jobs:
         ls -la /usr/lib/llvm-18/lib/libclang.so
         which llvm-config-18
 
-        # Build OpenCV 4.12 from source since Ubuntu 24.04 only has 4.6
-        echo "Building OpenCV 4.12 from source..."
+        # Build OpenCV 4.11 from source since Ubuntu 24.04 only has 4.6
+        echo "Building OpenCV 4.11 from source..."
         cd /tmp
-        wget -q https://github.com/opencv/opencv/archive/4.12.0.tar.gz
-        tar -xzf 4.12.0.tar.gz
-        cd opencv-4.12.0
+        wget -q https://github.com/opencv/opencv/archive/4.11.0.tar.gz
+        tar -xzf 4.11.0.tar.gz
+        cd opencv-4.11.0
         mkdir build
         cd build
         cmake \
@@ -348,19 +348,19 @@ jobs:
     - name: Install maturin (Linux/macOS)
       run: pip install maturin[patchelf]
 
-    - name: Install OpenCV (macOS)
+    - name: Install OpenCV build dependencies (macOS)
       if: runner.os == 'macOS'
       run: |
-        # Install OpenCV and LLVM via Homebrew
-        brew install opencv llvm
+        # Build against a pinned OpenCV version instead of Homebrew's moving target
+        brew install llvm cmake nasm pkg-config
         # Set up LLVM environment for opencv-rust bindings (keg-only installation)
         echo "LIBCLANG_PATH=/opt/homebrew/opt/llvm/lib" >> $GITHUB_ENV
         echo "LLVM_CONFIG_PATH=/opt/homebrew/opt/llvm/bin/llvm-config" >> $GITHUB_ENV
         echo "DYLD_LIBRARY_PATH=/opt/homebrew/opt/llvm/lib:$DYLD_LIBRARY_PATH" >> $GITHUB_ENV
-        # Set up OpenCV environment
-        echo "OPENCV_LINK_LIBS=opencv_calib3d,opencv_core,opencv_dnn,opencv_features2d,opencv_flann,opencv_highgui,opencv_imgcodecs,opencv_imgproc,opencv_ml,opencv_objdetect,opencv_photo,opencv_stitching,opencv_video,opencv_videoio" >> $GITHUB_ENV
-        echo "OPENCV_LINK_PATHS=/opt/homebrew/lib" >> $GITHUB_ENV
-        echo "OPENCV_INCLUDE_PATHS=/opt/homebrew/include/opencv4" >> $GITHUB_ENV
+
+    - name: Build static OpenCV (macOS)
+      if: runner.os == 'macOS'
+      run: ./scripts/build-opencv-static.sh
 
     - name: Build wheels (Linux)
       if: runner.os == 'Linux'
@@ -369,6 +369,11 @@ jobs:
     - name: Build wheels (macOS)
       if: runner.os == 'macOS'
       run: maturin build --release --out dist --no-default-features --features python-bindings,simd,opencv,metal
+      env:
+        OPENCV_INCLUDE_PATHS: ${{ github.workspace }}/third_party/opencv-static/include/opencv4
+        OPENCV_LINK_PATHS: ${{ github.workspace }}/third_party/opencv-static/lib
+        OPENCV_LINK_LIBS: static=opencv_world,static=avformat,static=avcodec,static=avfilter,static=swresample,static=swscale,static=avutil,static=jpeg,static=png,static=tiff,static=webp,static=jasper,framework=Accelerate,dylib=c++,framework=OpenCL,z
+        OPENCV_DISABLE_PROBES: pkg_config,cmake,vcpkg,vcpkg_cmake
 
 
     - name: Upload wheels
@@ -378,79 +383,86 @@ jobs:
         path: dist/
 
   build-wheels-aarch64:
-    name: Build wheels (Linux aarch64)
+    name: Build wheels (Linux aarch64 / DGX Spark)
     runs-on: ubuntu-24.04-arm
     strategy:
       matrix:
-        python-version: ["3.11", "3.12", "3.13"]
+        python: ["3.11", "3.12", "3.13"]
 
     steps:
     - uses: actions/checkout@v4
 
-    - name: Install OpenCV and LLVM dependencies
+    - name: Set up Python ${{ matrix.python }}
+      uses: actions/setup-python@v5
+      with:
+        python-version: ${{ matrix.python }}
+
+    - name: Cache static OpenCV
+      uses: actions/cache@v4
+      with:
+        path: third_party/opencv-static
+        key: opencv-static-4.11.0-codecs-jasper-ffmpeg-no-itt-no-openjpeg-linux-aarch64-${{ runner.arch }}-manylinux_2_39
+
+    - name: Install build dependencies
       run: |
+        export DEBIAN_FRONTEND=noninteractive
+
         sudo apt-get update
-        sudo apt-get install -y pkg-config clang libclang-dev llvm-dev wget cmake build-essential
-        # Install LLVM 18 which has the proper libclang.so symlink
+        sudo apt-get install -y pkg-config clang libclang-dev llvm-dev wget cmake build-essential curl nasm yasm
+
         sudo apt-get install -y llvm-18 llvm-18-dev llvm-18-tools libclang1-18 libclang-cpp18
-        # Set environment variables for LLVM 18
-        echo "LIBCLANG_PATH=/usr/lib/llvm-18/lib" >> $GITHUB_ENV
         echo "LLVM_CONFIG_PATH=/usr/bin/llvm-config-18" >> $GITHUB_ENV
-        echo "LD_LIBRARY_PATH=/usr/lib/llvm-18/lib:$LD_LIBRARY_PATH" >> $GITHUB_ENV
-
-        # Build OpenCV 4.12 from source
-        echo "Building OpenCV 4.12 from source..."
-        cd /tmp
-        wget -q https://github.com/opencv/opencv/archive/4.12.0.tar.gz
-        tar -xzf 4.12.0.tar.gz
-        cd opencv-4.12.0
-        mkdir build
-        cd build
-        cmake \
-          -DCMAKE_BUILD_TYPE=RELEASE \
-          -DCMAKE_INSTALL_PREFIX=/usr/local \
-          -DWITH_CUDA=OFF \
-          -DWITH_OPENGL=OFF \
-          -DWITH_OPENCL=OFF \
-          -DWITH_TBB=OFF \
-          -DWITH_IPP=OFF \
-          -DWITH_1394=OFF \
-          -DWITH_V4L=OFF \
-          -DWITH_GTK=OFF \
-          -DBUILD_TESTS=OFF \
-          -DBUILD_PERF_TESTS=OFF \
-          -DBUILD_EXAMPLES=OFF \
-          -DBUILD_opencv_python2=OFF \
-          -DBUILD_opencv_python3=OFF \
-          -DBUILD_opencv_java=OFF \
-          -DOPENCV_GENERATE_PKGCONFIG=ON \
-          ..
-        make -j$(nproc)
-        sudo make install
-        sudo ldconfig
 
-        # Set OpenCV environment variables
-        echo "PKG_CONFIG_PATH=/usr/local/lib/pkgconfig:$PKG_CONFIG_PATH" >> $GITHUB_ENV
-        echo "LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH" >> $GITHUB_ENV
+        LIBCLANG_PATH="/usr/lib/llvm-18/lib"
+        if [ ! -f "$LIBCLANG_PATH/libclang.so" ]; then
+          LIBCLANG_PATH=$(find /usr/lib/llvm* -name "libclang.so" 2>/dev/null | head -1 | xargs dirname)
+        fi
 
-    - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v4
-      with:
-        python-version: ${{ matrix.python-version }}
-
-    - name: Install Rust toolchain
-      uses: dtolnay/rust-toolchain@master
-      with:
-        toolchain: 1.89
+        echo "LIBCLANG_PATH=$LIBCLANG_PATH" >> $GITHUB_ENV
+        echo "LD_LIBRARY_PATH=$LIBCLANG_PATH:$LD_LIBRARY_PATH" >> $GITHUB_ENV
+        echo "Using libclang.so at: $LIBCLANG_PATH/libclang.so"
 
-    - name: Install maturin
-      run: pip install maturin[patchelf]
+    - name: Build static OpenCV
+      run: ./scripts/build-opencv-static.sh
 
     - name: Build wheels
-      run: maturin build --release --out dist --no-default-features --features python-bindings,simd,opencv
+      uses: PyO3/maturin-action@v1
+      with:
+        target: aarch64
+        args: --release --out dist --no-default-features --features python-bindings,simd,opencv
+        sccache: 'false'
+        container: off
+        manylinux: manylinux_2_39
+      env:
+        PYO3_USE_ABI3_FORWARD_COMPATIBILITY: '1'
+        LIBCLANG_PATH: ${{ env.LIBCLANG_PATH }}
+        LLVM_CONFIG_PATH: ${{ env.LLVM_CONFIG_PATH }}
+        OPENCV_INCLUDE_PATHS: ${{ github.workspace }}/third_party/opencv-static/include/opencv4
+        OPENCV_LINK_PATHS: ${{ github.workspace }}/third_party/opencv-static/lib
+        OPENCV_LINK_LIBS: static=opencv_world,static=avformat,static=avcodec,static=avfilter,static=swresample,static=swscale,static=avutil,static=jpeg,static=png,static=tiff,static=webp,static=z,static=jasper,dylib=stdc++
+        OPENCV_DISABLE_PROBES: pkg_config,cmake,vcpkg,vcpkg_cmake
+
+    - name: Repair wheels
+      env:
+        OPENCV_LINK_PATHS: ${{ github.workspace }}/third_party/opencv-static/lib
+      run: ./scripts/repair-wheel.sh dist/*.whl
+
+    - name: Prune unrepaired wheels
+      run: find dist -type f -name '*.linux_*.whl' -print -delete
+
+    - name: Install built wheel
+      run: |
+        wheel="$(find dist -type f -name '*.whl' | sort | head -n 1)"
+        if [ -z "${wheel}" ]; then
+          echo "No wheel files found under dist/"
+          find dist -maxdepth 3 -type f || true
+          exit 1
+        fi
+        pip install "${wheel}"
+        python -c 'import trainingsample'
 
     - name: Upload wheels
       uses: actions/upload-artifact@v4
       with:
-        name: wheels-linux-aarch64-py${{ matrix.python-version }}
+        name: wheels-linux-aarch64-py${{ matrix.python }}
         path: dist/

{trainingsample-0.2.16 → trainingsample-0.2.18}/.github/workflows/publish.yml

@@ -49,7 +49,7 @@ jobs:
       uses: actions/cache@v4
       with:
         path: third_party/opencv-static
-        key: opencv-static-4.12.0-codecs-jasper-ffmpeg-no-itt-no-openjpeg-linux-${{ matrix.os }}-${{ runner.arch }}-${{ matrix.manylinux }}
+        key: opencv-static-4.11.0-codecs-jasper-ffmpeg-no-itt-no-openjpeg-linux-${{ matrix.os }}-${{ runner.arch }}-${{ matrix.manylinux }}
 
     - name: Install build dependencies
       run: |
@@ -112,7 +112,7 @@ jobs:
         # OpenCV static linking configuration
         OPENCV_INCLUDE_PATHS: ${{ github.workspace }}/third_party/opencv-static/include/opencv4
         OPENCV_LINK_PATHS: ${{ github.workspace }}/third_party/opencv-static/lib
-        OPENCV_LINK_LIBS: static=opencv_world,static=avformat,static=avcodec,static=avfilter,static=swresample,static=swscale,static=avutil,static=jpeg,static=png,static=tiff,static=webp,static=z,static=jasper,static=stdc++
+        OPENCV_LINK_LIBS: static=opencv_world,static=avformat,static=avcodec,static=avfilter,static=swresample,static=swscale,static=avutil,static=jpeg,static=png,static=tiff,static=webp,static=z,static=jasper,dylib=stdc++
         OPENCV_DISABLE_PROBES: pkg_config,cmake,vcpkg,vcpkg_cmake
 
     - name: Repair wheels
@@ -147,7 +147,7 @@ jobs:
       uses: actions/cache@v4
       with:
         path: third_party/opencv-static
-        key: opencv-static-4.12.0-codecs-jasper-ffmpeg-no-itt-no-openjpeg-linux-aarch64-${{ runner.arch }}-manylinux_2_39
+        key: opencv-static-4.11.0-codecs-jasper-ffmpeg-no-itt-no-openjpeg-linux-aarch64-${{ runner.arch }}-manylinux_2_39
 
     - name: Install build dependencies
       run: |
@@ -194,7 +194,7 @@ jobs:
         # OpenCV static linking configuration
         OPENCV_INCLUDE_PATHS: ${{ github.workspace }}/third_party/opencv-static/include/opencv4
         OPENCV_LINK_PATHS: ${{ github.workspace }}/third_party/opencv-static/lib
-        OPENCV_LINK_LIBS: static=opencv_world,static=avformat,static=avcodec,static=avfilter,static=swresample,static=swscale,static=avutil,static=jpeg,static=png,static=tiff,static=webp,static=z,static=jasper,static=stdc++
+        OPENCV_LINK_LIBS: static=opencv_world,static=avformat,static=avcodec,static=avfilter,static=swresample,static=swscale,static=avutil,static=jpeg,static=png,static=tiff,static=webp,static=z,static=jasper,dylib=stdc++
         OPENCV_DISABLE_PROBES: pkg_config,cmake,vcpkg,vcpkg_cmake
 
     - name: Repair wheels
@@ -229,7 +229,7 @@ jobs:
       uses: actions/cache@v4
       with:
         path: third_party/opencv-static
-        key: opencv-static-4.12.0-codecs-jasper-ffmpeg-no-itt-no-openjpeg-macos-${{ runner.os }}-${{ runner.arch }}
+        key: opencv-static-4.11.0-codecs-jasper-ffmpeg-no-itt-no-openjpeg-macos-${{ runner.os }}-${{ runner.arch }}
 
     - name: Install build dependencies
       run: |
@@ -372,7 +372,7 @@ jobs:
         fi
         echo "Installing wheel: ${wheel}"
         pip install "${wheel}"
-        python -c 'import trainingsample' || echo "Import failed"
+        python -c 'import trainingsample'
 
   release:
     name: Release

trainingsample-0.2.18/BENCHMARKS.md

@@ -0,0 +1,137 @@
+# Performance Benchmarks
+
+TrainingSample includes benchmarks for common preprocessing operations: crop, resize, luminance, resize-plus-luminance pipelines, and video frame resizing. The benchmarks are meant to catch regressions and provide workload-specific guidance, not to guarantee universal speedups over OpenCV or NumPy.
+
+## Running Benchmarks
+
+Use the repository virtual environment when available:
+
+```bash
+.venv/bin/python -m pytest tests/test_performance_benchmarks.py -q -s
+```
+
+To run every Python test and benchmark marker in the repo:
+
+```bash
+.venv/bin/python -m pytest -q
+```
+
+For a fresh source build before measuring:
+
+```bash
+env -u OPENCV_LINK_LIBS -u OPENCV_LINK_PATHS -u OPENCV_INCLUDE_PATHS \
+    -u LIBCLANG_PATH -u LLVM_CONFIG_PATH \
+    .venv/bin/maturin develop --release
+```
+
+The OpenCV Rust binding needs a discoverable OpenCV and Clang installation. On this development host, stale macOS-style OpenCV and LLVM environment variables had to be unset before the build could probe the system OpenCV installation.
+
+## Current Local Snapshot
+
+Last measured command:
+
+```bash
+.venv/bin/python -m pytest tests/test_performance_benchmarks.py -q -s
+```
+
+Environment:
+
+- Linux x86_64
+- CPython 3.13
+- NumPy 2.3.4
+- system OpenCV 4.11 via the Rust `opencv` crate
+- release build installed with `maturin develop --release`
+
+Point-in-time scenario timings from the benchmark output:
+
+| Scenario | Before optimization | After optimization | Comparison after optimization |
+|----------|---------------------|--------------------|-------------------------------|
+| Crop batch, 16 images | 22.9 ms | 0.4 ms | NumPy slicing was still faster because it returns views |
+| Mixed-shape crop, 8 images | 50.2 ms | 3.3 ms | NumPy slicing loop was near-zero because it returns views |
+| Resize, 4 mixed-size images | 4.1 ms | 0.4 ms | OpenCV loop: 2.6 ms |
+| Luminance, 4 mixed-size images | 10.4 ms | 0.6 ms | OpenCV loop: 0.9 ms |
+| Resize + luminance pipeline, 4 images | 5.9 ms | 0.6 ms | OpenCV loop: 2.1 ms |
+| Mixed-shape luminance, 6 images | 78.3 ms | 3.3 ms | NumPy loop: 19.4 ms |
+
+Pytest-benchmark means from the same focused run:
+
+| Benchmark | Mean |
+|-----------|------|
+| Center crop | 55.2 us |
+| Resize operations | 353.1 us |
+| Luminance calculation | 417.2 us |
+| Crop operations | 583.8 us |
+| Pipeline | 3.44 ms |
+| Video processing | 2.85 ms |
+
+A full `pytest -q` run also passed and produced similar benchmark ordering, with normal run-to-run variance.
+
+## What Changed in the Latest Optimization
+
+- Owned Rust `ndarray` outputs are transferred into NumPy with `from_owned_array_bound`, avoiding an additional copy in Python-facing result conversion.
+- Contiguous luminance inputs use a channel-sum fast path. Instead of computing weighted luminance per pixel, it sums R, G, and B separately and applies the weights once at the end.
+- Non-contiguous arrays still use the general ndarray path for correctness.
+
+## Benchmark Categories
+
+### Image Operations
+
+- `batch_crop_images`
+- `batch_center_crop_images`
+- `batch_random_crop_images`
+- `batch_resize_images`
+- `batch_calculate_luminance`
+
+### Pipeline Operations
+
+- resize followed by luminance
+- crop followed by resize
+- mixed input sizes and output sizes
+
+### Video Operations
+
+- `batch_resize_videos` with frame batches shaped `(T, H, W, 3)`
+
+## Interpreting Results
+
+Use these benchmarks to answer practical questions:
+
+- Is a change adding extra Rust-to-NumPy copies?
+- Are contiguous arrays staying on the fast path?
+- Is resize dominated by OpenCV work or Python binding overhead?
+- Does a mixed-shape batch still behave reasonably?
+- Is a video processing change accidentally introducing per-frame Python overhead?
+
+Some comparisons need context:
+
+- NumPy crop by slicing often returns a view, so it can be much faster than any function that returns owned cropped arrays.
+- Very small images can be dominated by Python call overhead.
+- Large images can be dominated by memory bandwidth rather than arithmetic.
+- OpenCV performance varies by build options, CPU features, and linked libraries.
+
+## Quality Checks
+
+The tests validate basic output behavior alongside timing:
+
+- Crop outputs have expected shape and match NumPy slicing where ownership differences do not matter.
+- Resize outputs have expected shape and are close to OpenCV output for the configured interpolation.
+- Luminance stays within a small tolerance of NumPy/OpenCV-style references.
+- Non-contiguous arrays are accepted by safe luminance paths and rejected by strict zero-copy crop/resize paths.
+
+## Regression Signals
+
+Investigate if a change causes:
+
+- Public batch crop to return to multi-millisecond timings for small batches.
+- Luminance on contiguous RGB arrays to lose the channel-sum fast path.
+- Resize benchmarks to add large overhead beyond OpenCV work.
+- Video resizing to scale with per-frame Python object churn.
+- Memory usage to grow unexpectedly for repeated batch calls.
+
+## Future Benchmark Work
+
+- Store historical benchmark results by commit and host.
+- Add explicit memory allocation tracking for Python-facing APIs.
+- Separate view-returning crop comparisons from owned-output crop comparisons.
+- Add more video pipeline benchmarks.
+- Document hardware and OpenCV build details in benchmark artifacts.