Blindness is a very common and unendurable disability among many disabilities. According to the World Health Organization(WHO), there are 285 million visually impaired people. The proposed iVision (Visually Impaired) Assistant System is developed to assist visually impaired people.
Our proposed system uses Tiny-YoloV3 using COCO dataset. At present it can detect 80 classes. We have used imageAI and openCV to detect, predict and pipeline the results to our frontend.
The Mean Average Precision (mAP@50) is 66.7
Device Features:
=> Detect People and Objects in front such as:
Vehicles
Animals
Obstructions in Streets
=> Audio alert for the object that is detected
Challenges faced:
=>Dependency Incompabilities.
=>Deploying the model.
=>Conversion to .tflite. mAP@50 - 66.7 is our current Mean Average Precision.
Languages and Tools:
To Run:
Step 1: pip instal -r requirement.txt
Step 2: wget https://drive.google.com/file/d/15YQqbp5iPCX3Rcvf7CFdZX_OfjuqfOiB/view?usp=sharing