The title of my project

Detection Sign Language phrases

Please watch the below video for a short demo

Data set collection and Process: I collected around 300 - 400 images for each expression. Then augmentation was done to increase the data to 700 images for each expression, which resulted in about 5500 images for the whole dataset. It's trained using the Efficientnet0 (TensorFlow) algorithm with batch size to 32 and number of steps to 40000 on Google colab GPU.

Result : Above machine learning model detects 9 different objects which are numbers one to four then "Thanks", "I love you", 'Yes' , 'No' , 'Hello'