Master different vision tasks with pre-trained Florence-2 | Community Q&A (Jul 3)

Master different vision tasks with pre-trained Florence-2 in this live coding community Q&A.
Resources:
- Roboflow: roboflow.com
- 📓 Florence notebook: colab.research.google.com/github/roboflow-ai/notebooks/blob/main/notebooks/how-to-finetune-florence-2-on-detection-dataset.ipynb
- 🗞 Florence-2 arXiv paper: arxiv.org/abs/2311.06242
- 🗞 Florence-2 overview blog post: blog.roboflow.com/florence-2
- 🗞 Florence-2 fine-tuning blog post: blog.roboflow.com/fine-tune-florence-2-object-detection
- 🔗 Florence-2 HF Space: huggingface.co/spaces/gokaygokay/Florence-2
Stay updated with the projects I'm working on at github.com/roboflow and github.com/SkalskiP! ⭐

[@adiasz]

Great content. Thanx!

[@AlbaraFawazShehadeh]

Great, we want tutorial how to fine tune Florence-2 for handwritten recognition task

[@JV-vp5vl]

This RF channel & blog really are the one and only best sources of information to stay on the edge for intermediate data scientists/ml engineers with a practical approach. Ultralytics YT channel is also improving a lot lately! I

[@Roboflow]

Thanks a lot! 🙏🏻 It so much pleasure to read comments like this.

[@JV-vp5vl]

@@Roboflow my pleasure! Will continue to ad you among my peers.

[@AlbaraFawazShehadeh]

Great, how to fine tune florence-2 for handwritten recognition with region tasks

[@user-xs7oy8hs5t]

Great job. I have a question. Could this model help me detect image manipulation traces, such as physical photo substitution or digital tampering like splicing, cloning, and airbrushing? Feel free to suggest any other model. Thank you

[@Roboflow]

Hi :) That unfortunately does not sound like a use case for Florence-2

[@user-xs7oy8hs5t]

​@@Roboflow Could you suggest a model that I can fine-tune for this problem?

[@Alan-b8o]

Hey, how does one use an exported model after training is complete on colab locally? Couldn't ask during the livestream

[@terranthegaint405]

provide the path to the saved model when loading it, use the code in the colab notebook he was showing as a reference.

[@Alan-b8o]

@@terranthegaint405 Could you show me a snippet of what part exactly do I need to alter?

[@terranthegaint405]

@@Alan-b8o CHECKPOINT = "microsoft/Florence-2-base-ft" REVISION = 'refs/pr/6' use the path to your model in place of "microsoft/Florence-2-base-ft" and remove REVISION and then modify model and processor to look like this model = AutoModelForCausalLM.from_pretrained(CHECKPOINT, trust_remote_code=True).to(DEVICE) processor = AutoProcessor.from_pretrained(CHECKPOINT, trust_remote_code=True)