WB1 Machine Vision and Its Applications -Detection and Tracking
Time : 15:20~16:50
Room : Sapphire		 Chair : Prof.Daijjin Kim (POSTECH, Korea)
15:20~15:50        WB1-1
Joint Person Re-identification and Camera Network Topology Inference in Camera Networks

Yeong-Jun Cho, Kuk-Jin Yoon(KAIST, Korea)
In this paper, we propose a scalable and automatic joint framework which solves both person re-identification and camera topology inference, iteratively. Experimental results using public and our person re-identification datasets show that the proposed methods are promising for person re-identification in the large-scale camera network.
15:50~16:20        WB1-2
High-speed High-performance Visual Tracker via Correlation Filter with Compressed Deep Feature

Jongwon Choi, Kyuewang Lee, Jiyeoup Jeong, Jin Young Choi(Seoul National University, Korea)
This paper introduces a context-aware correlation filter based tracker to achieve both high speed and high performance. We achieve high speed via deep feature compression based on a context-aware scheme utilizing multiple expert auto-encoders. To achieve high performance with the compressed feature map, we introduce extrinsic denoising processes and a new orthogonality loss term for pre-training and fine-tuning of the expert auto-encoders. In experiments, the proposed tracker is verified to achieve a comparable performance to state-of-the-art with running at over 100 fps.
16:20~16:50        WB1-3
3D Point Cloud Classification with Deep Learning and Local Attention

Weonsuk Lee(POSTECH, Korea), Bohyung Han(Seoul National University, Korea)
We propose a simple neural network module for 3D point cloud classification to exploit local information. Our module computes an attention for each point pair using a Gaussian of Euclidean distance in 3D space. The features are then aggregated locally through the computed attentions. We show that our algorithm achieves competitive accuracy on ModelNet40 benchmark.

<<   1   >>