A framework for automatically constructing a dataset for training a vehicle detector. (18th March 2019)

• Record Type: Journal Article
• Authors: Kim, Changyon; Gwak, Jeonghwan; Shim, Daeyoung; Jeon, Moongu
• Is Part Of: International journal of computational vision and robotics. Volume 9:Number 2(2019)
• Page Start: 192
• Page End: 206
• Publication Date: 2019-03-18
• Subjects: object detection -- optimal vehicle detector -- appearance model -- scene condition investigation -- automatic sample collection; Computer vision -- Periodicals; Robotics -- Periodicals; Artificial intelligence -- Periodicals; 006.3705
• Languages: English
• ISSNs: 1752-9131

Abstract

Object detection based on a trained detector has been widely applied to diverse tasks such as pedestrian, face, and vehicle detection. In such approach, detectors are learned offline with an enormous number of training samples. However, the approach has a significant drawback that heavy intervention and effort, as well as domain knowledge, of a human are essentially required to construct a reliable training dataset. To remedy this drawback, we propose a framework to collect and label training samples automatically. By analysing information of foreground blobs obtained from background subtraction results, a training dataset can be constructed without any human's effort. Also, condition investigation of scenes is performed periodically to check the suitability of sample candidates. As a result, it generates an accurate vehicle detector. With the proposed method, training samples can be automatically collected only when vehicle blobs in the given scene provide suitable appearance information. The effectiveness of the proposed framework is demonstrated from vehicle detection tasks under real traffic environments.