In unmanned, the traffic sign recognition is an important task. This project LISA US traffic signs dataset into training objects using YOLOv3 target detection method for real-time traffic sign recognition.
Specific projects include process includes: installation Darknet, download LISA traffic signs data, data sets, format conversion, to modify the configuration file, LISA training data set, testing the trained network model, performance statistics (mAP calculate and draw the PR curve) and first test box clustering.
YOLOv3 can be performed in real-time end-to-target detection, to fast reflexes. This course will teach you to use the hand YOLOv3 implement traffic sign recognition. YOLOv3 this course use Darknet, made presentations on the project Ubuntu. Darknet is the use of lightweight open source deep learning framework implemented in C, less dependent, portability is good, worthy of further study and exploration.
I introduced about YOLOv3 series of courses include:
(1) YOLOv3 target detection combat: training their own data sets
(2) YOLOv3 target detection combat: Traffic sign recognition (ie this course)
Course Link: https://edu.51cto.com/course/18279.html
(3) YOLOv3 target detection: Principles and analytical source
(4) YOLOv3 target detection: Improvement Network Model
Curriculum: compared to "YOLOv3 target detection combat training their own data set," The main difference between this course is to learn the data set format conversion, that LISA data set conversion from csv format to format and PASCAL VOC YOLO format YOLOv3 need. The course data set provides Python code format conversion.
Please everyone's attention over the course, and choose to study.
The figure is the result of the test conducted using YOLOv3 traffic sign recognition:
