Replies: 1 comment
-
Hi, changing the threshold by changing the config files is not possible, as the threshold is currently hard coded unfortunately. This parameter should belong in the model configs… When invoking the analyzer and suppose the table transformer model is the predictor of the first layout component you can get access to the parameter analyzer.pipeline_component_list[0].predictor.config.threshold |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi all,
Since deepdoctection uses Table Transformer, what config setting corresponds to the "threshold" in the following HuggingFace code? Here, when I reduce the threshold, I get more tables that have low confidence.
image = Image.open(file_path).convert("RGB")
image_processor = AutoImageProcessor.from_pretrained("microsoft/table-transformer-detection")
...
results = image_processor.post_process_object_detection(outputs, threshold=0.2, target_sizes=target_sizes)[0]
Thanks,
Krishna
Beta Was this translation helpful? Give feedback.
All reactions