Skip to content

Exclude classes from inference using pretrained or custom models #2073

Exclude classes from inference using pretrained or custom models

Exclude classes from inference using pretrained or custom models #2073

Workflow file for this run

name: CI (PyTorch 1.12.1, TorchVision 0.13.1)
on:
push:
branches: [main]
paths-ignore:
- '**.md'
- '**.ipynb'
- '**.cff'
pull_request:
branches: [main]
paths-ignore:
- '**.md'
- '**.ipynb'
- '**.cff'
jobs:
build:
runs-on: ubuntu-latest
strategy:
matrix:
operating-system: [ubuntu-latest, windows-latest, macos-latest]
python-version: [3.8, 3.9, '3.10']
fail-fast: false
steps:
- name: Checkout
uses: actions/checkout@v4
- name: Set up Python
uses: actions/setup-python@v5
with:
python-version: ${{ matrix.python-version }}
- name: Restore Ubuntu cache
uses: actions/cache@v4
if: matrix.operating-system == 'ubuntu-latest'
with:
path: ~/.cache/pip
key: ${{ matrix.os }}-${{ matrix.python-version }}-${{ hashFiles('**/setup.py')}}
restore-keys: ${{ matrix.os }}-${{ matrix.python-version }}-
- name: Restore MacOS cache
uses: actions/cache@v4
if: matrix.operating-system == 'macos-latest'
with:
path: ~/Library/Caches/pip
key: ${{ matrix.os }}-${{ matrix.python-version }}-${{ hashFiles('**/setup.py')}}
restore-keys: ${{ matrix.os }}-${{ matrix.python-version }}-
- name: Restore Windows cache
uses: actions/cache@v4
if: matrix.operating-system == 'windows-latest'
with:
path: ~\AppData\Local\pip\Cache
key: ${{ matrix.os }}-${{ matrix.python-version }}-${{ hashFiles('**/setup.py')}}
restore-keys: ${{ matrix.os }}-${{ matrix.python-version }}-
- name: Update pip
run: python -m pip install --upgrade pip
- name: Lint with flake8, black and isort
run: |
pip install -e .[dev]
# stop the build if there are Python syntax errors or undefined names
python -m scripts.run_code_style check
# exit-zero treats all errors as warnings. The GitHub editor is 127 chars wide
flake8 . --exit-zero --max-complexity=10 --max-line-length=127
- name: Install core dependencies
run: >
pip install -r requirements.txt
- name: Install PyTorch(1.13.1) and TorchVision(0.14.1) on Linux and Windows
if: >
matrix.operating-system == 'ubuntu-latest' ||
matrix.operating-system == 'windows-latest'
run: >
pip install torch==1.13.1+cpu torchvision==0.14.1+cpu
-f https://download.pytorch.org/whl/torch_stable.html
- name: Install PyTorch on MacOS
if: matrix.operating-system == 'macos-latest'
run: pip install torch==1.13.1 torchvision==0.14.1
- name: Install MMDetection(3.0.0) with MMCV(2.0.0)
run: |
pip install mmengine==0.7.3
pip install mmcv==2.0.0 -f https://download.openmmlab.com/mmcv/dist/cpu/torch1.13.0/index.html
pip install mmdet==3.0.0
- name: Install YOLOv5(7.0.13)
run: >
pip install yolov5==7.0.13
- name: Install DeepSparse
run: >
pip install deepsparse
- name: Install Transformers(4.35.0)
run: >
pip install transformers==4.35.0
- name: Install pycocotools(2.0.7)
run: >
pip install pycocotools==2.0.7
- name: Install ultralytics(8.3.50)
run: >
pip install ultralytics==8.3.50
- name: Unittest for SAHI+YOLO11/RTDETR/MMDET/HuggingFace/Torchvision on all platforms
run: |
python -m unittest
- name: Install SAHI package from local setup.py
run: >
pip install -e .
- name: Test SAHI CLI
run: |
# help
sahi --help
# predict mmdet
sahi predict --source tests/data/ --novisual --model_path tests/data/models/mmdet/yolox/yolox_tiny_8x8_300e_coco_20211124_171234-b4047906.pth --model_config_path tests/data/models/mmdet/yolox/yolox_tiny_8xb8-300e_coco.py --image_size 320
sahi predict --source tests/data/coco_utils/terrain1.jpg --export_pickle --export_crop --model_path tests/data/models/mmdet/yolox/yolox_tiny_8x8_300e_coco_20211124_171234-b4047906.pth --model_config_path tests/data/models/mmdet/yolox/yolox_tiny_8xb8-300e_coco.py --image_size 320
sahi predict --source tests/data/coco_utils/ --novisual --dataset_json_path tests/data/coco_utils/combined_coco.json --model_path tests/data/models/mmdet/yolox/yolox_tiny_8x8_300e_coco_20211124_171234-b4047906.pth --model_config_path tests/data/models/mmdet/yolox/yolox_tiny_8xb8-300e_coco.py --image_size 320
# predict yolov5
sahi predict --no_sliced_prediction --model_type yolov5 --source tests/data/coco_utils/terrain1.jpg --novisual --model_path tests/data/models/yolov5/yolov5s6.pt --image_size 320
sahi predict --model_type yolov5 --source tests/data/ --novisual --model_path tests/data/models/yolov5/yolov5s6.pt --image_size 320
sahi predict --model_type yolov5 --source tests/data/coco_utils/terrain1.jpg --export_pickle --export_crop --model_path tests/data/models/yolov5/yolov5s6.pt --image_size 320
sahi predict --model_type yolov5 --source tests/data/coco_utils/ --novisual --dataset_json_path tests/data/coco_utils/combined_coco.json --model_path tests/data/models/yolov5/yolov5s6.pt --image_size 320
# coco yolov5
sahi coco yolov5 --image_dir tests/data/coco_utils/ --dataset_json_path tests/data/coco_utils/combined_coco.json --train_split 0.9
# coco evaluate
sahi coco evaluate --dataset_json_path tests/data/coco_evaluate/dataset.json --result_json_path tests/data/coco_evaluate/result.json
# coco analyse
sahi coco analyse --dataset_json_path tests/data/coco_evaluate/dataset.json --result_json_path tests/data/coco_evaluate/result.json --out_dir tests/data/coco_evaluate/