Object detection for video surveillance using the SSD approach
Thesis title in Czech: | Detekce objektů pro kamerový dohled pomocí SSD přístupu |
---|---|
Thesis title in English: | Object detection for video surveillance using the SSD approach |
Key words: | detekce objektů, kamerový dohled, hluboké neuronové sítě, architektura SSD |
English key words: | object detection, video surveillance, deep neural networks, SSD architecture |
Academic year of topic announcement: | 2018/2019 |
Thesis type: | diploma thesis |
Thesis language: | angličtina |
Department: | Department of Software Engineering (32-KSI) |
Supervisor: | doc. RNDr. Jakub Lokoč, Ph.D. |
Author: | hidden - assigned and confirmed by the Study Dept. |
Date of registration: | 13.02.2019 |
Date of assignment: | 13.02.2019 |
Confirmed by Study dept. on: | 21.02.2019 |
Date and time of defence: | 10.06.2019 09:00 |
Date of electronic submission: | 07.05.2019 |
Date of submission of printed version: | 10.05.2019 |
Date of proceeded defence: | 10.06.2019 |
Opponents: | RNDr. Petr Božovský, CSc. |
Guidelines |
Video surveillance has become an essential tool for monitoring human activities for purposes of security, market research, traffic management, etc. A constant need for observing and analyzing the growing number of video streams has created a demand for real-time automated systems for this task. Automated object detection delivers a possibility to use computing power for video analysis and provide notifications and summaries of interesting activities.
In this thesis, the student will examine and comprehensibly outline techniques used for object detection, focusing on approaches inspired by the Single Shot Detector (SSD). Designed detectors will focus on video surveillance, thus primarily detecting objects like people and vehicles. Specifically, the thesis will investigate multiple state-of-the-art image classification models in combination with the SSD approach for the bounding box prediction. The second goal is to identify a variant capable of real-time video analysis. The student will also analyze the possibilities of detector optimization for video surveillance purposes. |
References |
Dalal, Navneet, and Bill Triggs. "Histograms of oriented gradients for human detection." Computer Vision and Pattern Recognition, 2005. CVPR 2005. IEEE Computer Society Conference on. Vol. 1. IEEE, 2005.
Liu, Wei, et al. "SSD: Single shot multibox detector." European conference on computer vision (ECCV). LNCS, volume 9905, Springer, Cham, 2016. Goodfellow, Ian, et al. Deep learning. Vol. 1. Cambridge: MIT press, 2016. He, Kaiming, et al. "Deep residual learning for image recognition." Proceedings of the IEEE conference on computer vision and pattern recognition. Las Vegas, NV, 2016, pp. 770-778. doi: 10.1109/CVPR.2016.90. Huang, Jonathan, et al. "Speed/accuracy trade-offs for modern convolutional object detectors." IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 3296-3297, doi: 10.1109/CVPR.2017.351 |