Voice Assisted Traffic Sign Detection

It uses state-of-the-art object detection algorithm (YOLO) to identify regulatory and warning traffic signs. The dataset is a collection of images with different classes found across the internet. We used browser extension to gather images from several image hosting sites and also utilized road trip videos to gather field data.

Team Members

• Baltazar, Zeus James
• Bantuas, Junaid
• Basbacio, Martin Lorenzo
• Marquez, Ian Gabriel

Tools and Dependencies

• LabelImg by Tzutalin
• OpenCV
• Numpy
• Darknet by AlexeyAB

Classes with mAP

• stop (0.0%)
• intersection (0.0%)
• no u turn (0.0%)
• no left turn (0.0%)
• no right turn (0.0%)
• speed limit (0.0%)
• pedestrian (0.0%)
• warning (0.0%)
• no parking (0.0%)
• one way (0.0%)
• no entry (0.0%)

Training

The training was performed with the help of google colaboratory's free GPU and darknet neural network framework.