Visual attribute detction for pedestrian detection
Multimedia Tools and Applications(2016)
摘要
Attributes are expected to narrow down the semantic gap between low-level visual features and high-level semantic meanings. Such superiority motivates us to explore pedestrian attributes which has became a critical problem to boost image understanding and improve the performance of pedestrian detection, retrieval, re-identification, etc. Based on the PETA dateset, we manually relabel two subset VIPeR and PRID as our experimental dataset. Moreover, we proposed an evaluation protocol for researchers to evaluate pedestrian attribute classification algorithms. In this paper, we utilized two baseline methods to to demonstrate the performance of the attribute in pedestrian detection. The first one directly uses color and texture features to train Support Vector Machine (SVM) classification while the other one uses DSIFT (Dense SIFT) with Bag-of-Visual-Words (BoVW) to train SVM classification. Finally, we report and discuss the baseline performance on the database following the proposed evaluation protocol.
更多查看译文
关键词
Pedestrian attribute classification, Bag-of-Visual-Words, Support vector machine
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络