Same question. It would be super helpful if we could have the ground-truth of the bounding box of each cell in tables, by modifying your data processing code. Maybe in a similar way to your detection for table boundary? (add bounding box to the cell and check the pixel difference of the compiled document image)
Same question. It would be super helpful if we could have the ground-truth of the bounding box of each cell in tables, by modifying your data processing code. Maybe in a similar way to your detection for table boundary? (add bounding box to the cell and check the pixel difference of the compiled document image)