AbhishekSinghDhadwal/Indian-Sign-Language-Translator — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a Python app that translates Indian Sign Language from a webcam feed. I want it to feel close to real time, show the live camera view, and display the recognized sign clearly on screen.

Please make it work even when the background or lighting is not perfect. Use face detection as the trigger so the translator starts only when a person is actually in front of the camera. Then detect the hand, crop the hand area, clean up the image with skin segmentation, and send it to a trained sign recognition model. The model only needs to support the 10 sign classes from the dataset.

Also include the supporting scripts for preparing images, cleaning and augmenting the dataset, and training or saving the model so I can improve it later. Keep the main app easy to run from one file, with clear setup instructions, requirements, and notes about where model files or detection weights should go. Look up current docs online if needed.

Want more depth? Deep Reverse