shag527/Indian-Sign-Language-Recognition — reverse-engineered prompt

Build me a Python desktop app for Indian Sign Language recognition like this project describes. It should help speech or hearing impaired users communicate by recognizing hand signs for A to Z and 0 to 9 from a webcam or captured images, then showing the predicted letter or digit as text and speaking it out loud.

I also want the reverse mode, where a user can speak a word and the app shows the matching sign images in sequence. Include a simple login or account screen, then a main window with clear buttons for recognizing signs, creating or collecting new sign images, and using speech to signs.

Use the existing approach from the repo, with image preprocessing, hand segmentation, visual features, saved classifiers, and a CNN option if available. Please wire it up so it can run locally with the provided dataset and requirements, and add clear setup and run instructions. If anything is outdated, look up current docs online and make the smallest practical updates so it works.