ocrd_detectron2
ocrd_detectron2 copied to clipboard
bertsky
OCR-D wrapper for detectron2 based segmentation models
#2 I dont know exactly if this is what you are expecting, but I tested it with RGB Images and was able to process those, even with postprocessing.
Hi I am trying to run this library but no luck till now. I've followed the readme and I got deps properly installed. I am trying to run the model...
For example, in … |raw predictions|NMS result| |---|---| ||| … a 57% cross-column region does not get suppressed by two (mutually non-overlapping) column regions with 91% and 74%.
Since Detectron2 based models for textline segmentation have appeared, we should support them by adding a `operation_level=region`. Also, the `cat2class` mapping must be extended by `TextLine` type to allow its...
We currently only use Detectron2's `DefaultPredictor` for inference: https://github.com/bertsky/ocrd_detectron2/blob/0272d95a930d5136bba29e530a3530c13ab17166/ocrd_detectron2/segment.py#L126 But the [documentation](https://detectron2.readthedocs.io/en/latest/modules/engine.html?highlight=defaultpredictor#detectron2.engine.defaults.DefaultPredictor) says: > This is meant for simple demo purposes, so it does the above steps automatically. This is...
Build fails (and breaks ocrd_all builds, too). This is the failing part: ``` pip3 install --no-build-isolation -f "https://dl.fbaipublicfiles.com/detectron2/wheels/$CUDA/torch1.10/index.html" "git+https://github.com/facebookresearch/[email protected]#egg=detectron2" Looking in links: https://dl.fbaipublicfiles.com/detectron2/wheels/cu122/torch1.10/index.html Collecting detectron2 Cloning https://github.com/facebookresearch/detectron2 (to revision v0.6)...