https://github.com/starwit/sae-object-detector
SAE component that detects objects. It uses Yolo for object detection and can run individual models
https://github.com/starwit/sae-object-detector
ai computervision sae
Last synced: 5 months ago
JSON representation
SAE component that detects objects. It uses Yolo for object detection and can run individual models
- Host: GitHub
- URL: https://github.com/starwit/sae-object-detector
- Owner: starwit
- License: agpl-3.0
- Created: 2023-05-10T14:19:07.000Z (about 3 years ago)
- Default Branch: main
- Last Pushed: 2025-12-08T12:34:58.000Z (6 months ago)
- Last Synced: 2025-12-09T00:12:06.215Z (6 months ago)
- Topics: ai, computervision, sae
- Language: Python
- Homepage:
- Size: 441 KB
- Stars: 1
- Watchers: 3
- Forks: 1
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Object Detector
This component is part of the Starwit Awareness Engine (SAE). See umbrella repo here: https://github.com/starwit/starwit-awareness-engine
## How to set up
- Make sure you have Poetry installed (otherwise see here: https://python-poetry.org/docs/#installing-with-the-official-installer)
- Set environment variable `NVIDIA_TENSORRT_DISABLE_INTERNAL_PIP=True` (otherwise `tensorrt-*` installation will fail)
- Run `poetry install`
## How to Build
See [dev readme](doc/DEV_README.md) for build instructions.
## Model Development
In order to detect objects a trained model is necessary. See [model development](doc/Model_Development.md) documentation for more details.
## Github Workflows and Versioning
The following Github Actions are available:
* [PR build](.github/workflows/pr-build.yml): Builds python project for each pull request to main branch. `poetry install` and `poetry run pytest` are executed, to compile and test python code.
* [Build and publish latest image](.github/workflows/build-publish-latest.yml): Manually executed action. Same like PR build. Additionally puts latest docker image to internal docker registry.
* [Create release](.github/workflows/create-release.yml): Manually executed action. Creates a github release with tag, docker image in internal docker registry, helm chart in chartmuseum by using and incrementing the version in pyproject.toml. Poetry is updating to next version by using "patch, minor and major" keywords. If you want to change to non-incremental version, set version in directly in pyproject.toml and execute create release afterwards.
## Dependabot Version Update
With [dependabot.yml](.github/dependabot.yml) a scheduled version update via Dependabot is configured. Dependabot creates a pull request if newer versions are available and the compilation is checked via PR build.
## Changelog
### 4.2.1
- Update `sae-cv-base` to 0.2.0 (relevant for docker build)
- Update `openvino` to 2025.4.1
### 4.0.0
- Explicitly support PyTorch (on CPU, Nvidia and Intel GPU), TensorRT (on Nvidia) and OpenVINO (on Intel) versions of YOLO models
- Update CUDA libs -> TensorRT recompilation is needed!
- Improve error messages regarding model configuration
- Breaking config changes
- Move `inference_size` and `classes` into `.model`
- Rename `fp16_quantization` to `fp16`