TensorFlowLiteAPI
API for Image Classification based on https://lobe.ai
use your own model by binding the folder /api/models from docker
inspired by: https://github.com/jomjol/docker-synology-opencv-tensorflow-lite
TensorFLowLite without support the avx extension, can run on synology with intel processor
http://127.0.0.1:8000/redoc - api docs
http://127.0.0.1:8000/process - api
uvicorn server:app - start server
basic functionality:
GET /process query parameters:
- [Optional] model - name of the folder that contains the model
- [Required] url - the URL of the image to be processed
- [Optional] token - in case a token is needed in order to access the image
Output - JSON:
- url - the URL that was processed
- name - the temporary name of the processed file
- result:
- Prediction - the result of the process
- Confidences - accuracy of the result in float32