Open Vision API is an open source computer vision API based on open source models.
The following instructions detail how to set up the ova-server.
For information regarding the official clients and a quick demo of the API functionality head to:
- ova-client: the official Python client.
- ova: the official Rust client.
Make sure you have:
Set up a local environment using TensorFlow Lite as the backend framework.
$ just setup-tensorflow-liteSee documentation for a list of supported deep learning frameworks.
Download the models.
$ poetry run ./cli.py download --model=yolov4 --framework=tensorflow_lite --hardware=cpuRun the ova-server.
$ just run-with-tensorflow-lite
INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
INFO: Started server process [3588600]
INFO: Waiting for application startup.
INFO: Application startup complete.
INFO: Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)Get the official client.
$ git clone https://github.com/openvisionapi/ova-client
$ cd ova-client
$ just setup
$ OVA_DETECTION_URL=http://localhost:8000/api/v1/detection poetry run ./ova.py detection images/cat.jpegFor more information about the python client, please visit https://github.com/openvisionapi/ova-client
$ git clone https://github.com/openvisionapi/ova
$ cd ova
$ OVA_DETECTION_URL=http://localhost:8000/api/v1/detection cargo run -- detection -i assets/cat.jpegFor more information about the rust client, please visit https://github.com/openvisionapi/ova
All contributions are welcome!
To set up the development environment, simply run the command:
$ just devruff and mypy are used to ensure that contributions are stylized in a uniform manner.
To run the tests, simply use the commands:
$ just dev
$ just testThe complete documentation can be found here documentation
AGPLv3
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent