Is it possible to further train on a pre-trained model that I get from Roboflow Universe? For example, if I wanted to train “People Detection (general) Dataset” from the Roboflow Universe on additional images so it detects more of what I want it to detect, is that possible?
I think @Lenny gave a good answer on this one. To reiterate, yes, you can fine-tune and further train a Universe pre-trained model. Here are the steps:
- “Star” the model you’d like to fine-tune
- When you train a model, under Train from Public Checkpoint, select the model you want to train from
Right now, my project is completely blank but once I add training images (the one I want to train the Roboflow Universe API model on), the option to train from a public checkpoint will appear?
Once you generate a version and try to train, yes it should appear.
I’m not completely confident I understood you right, so if you’re not seeing something you should be, could you attach a screenshot to help us troubleshoot that for you?