How to implement active learning in local environment using docker image?

  • Project Type: Object detection and tracking
  • Operating System & Browser: macOS/ Google Chrome

I have created a model and downloaded the docker image over my local. I wanted to know how to implement active learning after doing these two steps. I tried running the inference over local and was successful in doing that but I am not able to understand the active learning pipeline. Can anyone please help out on this, I am a newbie on the Roboflow platform.


Hi @Kapil_Raj

Primarily, Roboflow is mostly in the cloud. You can annotate your images, manage your datasets, train your models, and infer on them in the cloud. When using local inference, it just shifts the inference portion locally, but nothing else.

Here’s a great article on active learning if you’d like to learn more: What is Active Learning?

You can use our image upload API to upload low confidence or incorrect images back into your project, where the model can learn from the new real-world data.