Reference: - 01-Design-Doc

Version 1

POST /api/run

• Input - Request body: {file: File, model_id: 'yolov8n'}
• Get the file and model id
• Looks up model.json to find the file path which belongs to the model id
• Loads the .pt file for that model
• Gets the image/video from the file path
• Runs the inference
• Return results as JSON

Now, update the run-detect logic according to this. and add a model.json file to store the model id and model path pair.

GET /api/models

• Returns a list of models in json format
• .json file should include: id, model name, path

POST /api/models

• (User uploads a new .pt file)
• Save the file to a path (backend/app/models/{model-name})
• Update models.json (id, model name, path)

DELETE /api/models/:id

• Update models.json

Should have two blueprints:

• model
• inference

Version 2