Hi! This program is quite impressive! I am enjoying using it to try to recognize various license plates. However, I've had a few issues in trying to increase the accuracy of the recognition. The main thing is, when I've tried training using different files, the classifications and flattened_images files that I use in the main.py file seem to cause the program to only recognize numbers!
I trained extensively using letters/different fonts, and I also compared the new generated classifications file to the original one, and it seems like the generated classifications file is much shorter than the one in the original repo. I was wondering if you knew what might be causing that and how to remedy it to recognize letters as well? Thanks!
Edit: The way I trained on the new file was to press the corresponding keys when the box was around the letters, and when it was around the state symbol, I just pressed enter. Hope this is helpful in fixing this issue!
Hi! This program is quite impressive! I am enjoying using it to try to recognize various license plates. However, I've had a few issues in trying to increase the accuracy of the recognition. The main thing is, when I've tried training using different files, the classifications and flattened_images files that I use in the main.py file seem to cause the program to only recognize numbers!
I trained extensively using letters/different fonts, and I also compared the new generated classifications file to the original one, and it seems like the generated classifications file is much shorter than the one in the original repo. I was wondering if you knew what might be causing that and how to remedy it to recognize letters as well? Thanks!
Edit: The way I trained on the new file was to press the corresponding keys when the box was around the letters, and when it was around the state symbol, I just pressed enter. Hope this is helpful in fixing this issue!