Vision & Learning Laboratory

Multi-Object Tracking

The goal of multi-object tracking is to estimate the states of multiple objects, such as locations, velocities, and sizes, while conserving their identifications. We developed confidence-based tracking and online appearance learning, and deep learning algorithms to solve this problem.

Publication

- Seong-Ho Lee, Myung-Yun Kim, and Seung-Hwan Bae, IEEE Access, 2018
- Seung-Hwan Bae, and Kuk-Jin Yoon, TPAMI, 2018
- Seung-Hwan Bae, and Kuk-Jin Yoon, CVPR, 2014
- Seung-Hwan Bae, and Kuk-Jin Yoon, TIP, 2014
[DEMO / CODE]

Object Detection & Deep Learning

An object detection is to predict object hypothesis of one or more classes given an image. We developed deep learning-based object detectors for large category object detection.

Achievement

- Seung-Hwan Bae, AAAI, 2019
- Winning the 2nd place at Large Scale Visual Recognition Challange 2017 (ILSVRC2017), Team Name: DeepView [Link]

Generative Adversarial Learning

Image generation and editing with generative adversarial learning.

Publication

- Jae-Yong Baek, Yong-Sang Yoo, and Seung-Hwan Bae, IEEE Access, 2019

Radar-Based Object Tracking

Automated multi-object tracking with data association and track management in cluttered environment. We developed SNR estimation algorithms with amplitude using MAP and SMC methods.

Publication

- Seung-Hwan Bae, IET-RSN, 2019
- Seung-Hwan Bae, Jongyoul Park and Kuk-Jin Yoon, IET-RSN, 2017
- Seung-Hwan Bae, and Kuk-Jin Yoon, IET-RSN, 2012

Polyp Detection

Automated polyp detection is to find locations and sizes of polyps in endoscopic or colonoscopic images automatically. We proposed a data sampling-based boosting framework and discriminative feature learning for polyp detection.

Publication

- Seung-Hwan Bae, and Kuk-Jin Yoon, TMI, 2015
- Yeong-Jun Cho, Seung-Hwan Bae, and Kuk-Jin Yoon, JMBE, 2016
[DEMO]

Face Forensics

The goal of Face Forensic is to classify and detect manipulated images by the advanced graphic tool (e.g. Photoshop) and artificial intelligence (e.g. DeepFake). We developed a face forensic algorithm based on our novel generative adversarial ensemble learning to handle this problem.

Publication

- Jae-Yong Baek, Yong-Sang Yoo, and Seung-Hwan Bae* IEEE Access, 2020