Hello,
Thx again for sharing your work.
In the process_handler of the live text function, you break the lines into text chunks.
Is there any specific reason for that ? Why not simply return the whole line text ?
Indeed, with the smaller text chunks, one has to rebuild the whole line by analyzing the bounding boxes.
For example, recognizing a text with e.g. a title like "HELLO WORLD" returns 2 elements (with bboxes close to each other): "HELLO" and "WORLD".
If there is not specific technical limitation, then I would suggest at least to leave the choice and use an option like "greedy", defaulting to false. When true, it would not cut lines, and when false it would behave as now.
Opinion ?
Aurélien
Hello,
Thx again for sharing your work.
In the process_handler of the live text function, you break the lines into text chunks.
Is there any specific reason for that ? Why not simply return the whole line text ?
Indeed, with the smaller text chunks, one has to rebuild the whole line by analyzing the bounding boxes.
For example, recognizing a text with e.g. a title like "HELLO WORLD" returns 2 elements (with bboxes close to each other): "HELLO" and "WORLD".
If there is not specific technical limitation, then I would suggest at least to leave the choice and use an option like "greedy", defaulting to false. When true, it would not cut lines, and when false it would behave as now.
Opinion ?
Aurélien