Karthikeyu/Indian-sign-language-recognition — reverse-engineered prompt

Build me a Python project that can recognize Indian Sign Language gestures from hand images, especially the alphabet A to Z and numbers 1 to 9. I want it to work like a simple research demo, where I can capture my own gesture photos if needed, process the images, train or load a model, and then test a gesture to see the predicted letter or number.

Please use the approach from the original idea, skin or hand segmentation, SURF features, bag of visual words, K means, and an SVM classifier. Include the saved model workflow if possible, plus a way to show the confusion matrix and accuracy results. Keep the scripts easy to run in order, with clear instructions for where to put the dataset and what command to run next.

Use Python 2.7 and the OpenCV setup needed for SURF. If anything is outdated, look up the current docs online and make the smallest changes needed to get it running.