Challenge Task (1) – Traffic Sign Classification
Objectives
The goal of this challenge is to encourage AI enthusiast in Macao high schools to apply their AI skills to solve some real world problems in order to make our city smarter.
The major task in this challenge is to design an algorithm that can accurately recognize traffic sign from a photo. This skill is particularly useful in intelligent transportation systems (e.g., auto-driving and traffic accident analysis).
We expect students to learn the following skills in order to prepare high quality submission for the challenge.
- Basic Image Processing (e.g., feature extraction and image transformation)
- Basic Machine Learning Skills (e.g., classified and evaluation)
- Basic Learning Models (e.g., SVM and Deep Learning)
Data
Input:
Image path, e.g., /xxxyyyzzz/img/1235789.jpg
and the rectangle [x1 y1 x2 y2] that indicates the sign boundary of the photo, e.g.,
158 87 397 302
Output:
Traffic sign ID for each traffic sign image (the mapping between the IDs and traffic signs can be found in the provided data, dataset.zip on Github)
Evaluation
Accuracy = the number of correct answers / the total number of photos