Sorry if my question seem dumb. DeepStack makes it easy to register a face quite easily and fast with the Face Registration API. However, it seem like other objects doesn’t have API for object registration, rather it seem the method for object is via Custom model which involves training, which is quite a process and a bit cumbersome in a situation where few new objects will be registered every other day.
I’m just wondering, is there a limitation for not having Object Registration API, just as the Face Registration API?
Hi @eazuka thanks for creating this thread. This is something we are exploring and once we have a working API that allows you register a few images and create an API for an object, we will release a beta version.
Having said that, the accuracy of such API won’t be as accurate as the standard DeepStack detection API or a Custom trained model. It may get better with more work but can hardly be as accurate as the standard trained detection models.
You can create an issue in the DeepStack repository about this
Thanks for the feedback @OlafenwaMoses. I’m new and still learning and wrapping my head around ML/AI and DeepStack has helped a lot. Your feedback is well appreciated.