In the era of artificial intelligence lot of object detection devices are available for visually impaired people. When it comes to subject of dumb-blind people there is none of devices available in market. Personal assistive gadgets are not available to them; only communication devices such as MyVox and Sparsha are available. In this paper we are proposing a personal assistive device to help dumb-blind people using Machine learning and Python code. It consists of multiple in-built options like object detection and communication module. A machine learning objection detection model is used, that is trained using the CNN technique for object detection. It can detect multiple objects at same time and detected objects name is intimated to user by audio output. To make communication between users with other people we are using keypad by pressing the keypad we can replay the stored audio which can be used for speech conversation.

Raspberry Pi, Machine Learning, Text to Voice Conversion, VNC Viewer, Python