Ching Lik Teo

I am a Ph.D. student at Department of Computer Science at University of Maryland, College Park. I am also affiliated to the Computer Vision Lab at UMIACS. My advisors are Yiannis Aloimonos and Cornelia Fermüller.

I am interested in Computer Vision, Computational Linguistics and how these two emerging fields combine to solve difficult problems that model human understanding, attention and recognition. I am also interested in robotic perception, especially on how we can model Vision and Language on mobile active agents.

I am a recipient of the Qualcomm Innovation Fellowship 2011, the UMD CS Department Fellowship award and DSO National Laboratories Post-graduate Scholarship.

In 2014, I was accepted into the UMD Clark School's Future Faculty Program, that focuses on grooming future teaching faculties in Engineering and Computer Science.

I am one of the main organizers for the UMD Computer Vision Student Seminars (CVSS) series.

I serve as a reviewer for several Computer Vision and Robotics conferences: CVPR, ICCV, ECCV, ICRA and IROS and journals: T-PAMI, IJRR, CVIU and IMAVIS.

Here's my updated CV and my Google scholar profile.

Latest News

2015

• 2 Papers accepted at ICCV 2015.
• Paper accepted at CVPR 2015.
• Paper accepted at ICRA 2015.

2014

• Paper accepted for publication in the International Journal of Robotics Research.

Current Research

I am working on developing theories and algorithms that integrates Language with Vision beyond the semantic (label) level. This extends the previous work of using the "Cognitive Dialog" framework by considering the integration of vision and language across the entire spectrum of visual processes: from high-level (semantic) to low-level (signal). We are currently developing efficient methods for influencing low-level to mid-level visual processes: edge detection, contour grouping and segmentation, using high-level representations derived from language. More information can be found here.

Software and data

• Curved reflection symmetry detection and segmentation code described in our ICCV 2015 paper. See the project page for details.
• Border ownership assignment code described in our CVPR 2015 paper. See the project page for details.
• Gestalt-like features used for border ownership assignment described in our CVPR 2015 paper. See the project page for details.
• Fast parts based affordance detection code via structured random forests (SRF) described in our ICRA 2015 paper. See the project page for details.
• Contour-based object detection code using the torque shape-context described in our IJRR paper. See the project page for details.
• Modulated torque code described in our ICRA 2013 paper. See the project page for details.

• Symmetric segments dataset derived from the SYMMAX-300 dataset.
• Border ownership datasets with groundtruth derived from: a) Berkeley Segmentation dataset (BSDS) and b) NYU-Depth V2.
• RGB-D Part Affordance Dataset consisting of 105 kitchen, workshop and garden tools from 17 object categories. > 10,000 RGB-Depth frames with hand-annotated groundtruth of 7 affordance categories.
• UMD Hand Manipulation dataset consisting of 6 video sequences (~1500 frames each) of 3 human subjects constructing a wooden frame using 5 types of tools under clutter, occlusion and manipulation.
• UMD Olympic Sports Attributes dataset that contains 16 sports activities from 730 videos. Fully annotated with action attributes over bounding boxes.
• UMD Clutter Dataset that contains 3 RGB-Depth sequences (~1000 frames) with different amounts of clutter and occlusions, annotated with spatial location of 7 object categories.
• UMD Sushi-Making dataset consisting of synced RGB + MOCAP data of 4 actors performing various actions connected with making sushi.
• UMD-Telluride Kinect Dataset consisting of RGB-Depth Kinect data of 11 kitchen actions using different tools.
• POETICON video dataset consisting of 6 complex activities involving 2 persons. Fully annotated, and hand segmented.
• UMD Complex Activities. 16 sequences of synced Kinect and SR4000 data of manipulatory hand actions using a variety of tools and objects collected from 4 different actors.

We are actively using the Robot Operating System (ROS) to develop codes for the Pioneer 3-DX (previously Videre Erratic) robot platform. See our ROS hints to get acquainted. It contains useful README, links and sample ROS codes as well.

Refereed Publications

Ching L. Teo, Cornelia Fermüller, Yiannis Aloimonos. Detection and Segmentation of 2D Curved Reflection Symmetric Structures. IEEE International Conference on Computer Vision, ICCV. 2015.

Francisco Barranco*, Ching L. Teo*, Cornelia Fermüller, Yiannis Aloimonos. Contour Detection and Characterization for Asynchronous Event Sensors. IEEE International Conference on Computer Vision, ICCV. 2015. (* indicates equal contribution)

Ching L. Teo, Cornelia Fermüller, Yiannis Aloimonos. Fast 2D Border Ownership Assignment. IEEE Conference on Computer Vision and Pattern Recognition, CVPR. 2015. Oral

Austin Myers, Ching L. Teo, Cornelia Fermüller, Yiannis Aloimonos. Affordance Detection of Tool Parts from Geometric Features. IEEE International Conference on Robotics and Automation, ICRA. 2015.

Ching L. Teo, Cornelia Fermüller, Yiannis Aloimonos. A Gestaltist approach to contour-based object recognition: Combining bottom-up and top-down cues. The International Journal of Robotics Research, IJRR, vol 34(4-5):627--652, 2015. doi:10.1177/0278364914558493.

Xiaodong Yu, Ching L. Teo, Yezhou Yang, Cornelia Fermüller, Yiannis Aloimonos. Action Attribute Detection from Sports Videos with Contextual Constraints. British Machine Vision Conference, BMVC. 2013.

Ching L. Teo, Austin Myers, Cornelia Fermüller, Yiannis Aloimonos. Embedding High-Level Information into Low Level Vision: Efficient Object Search in Clutter. IEEE International Conference on Robotics and Automation, ICRA. 2013. Oral

Yezhou Yang, Ching L. Teo, Cornelia Fermüller, Yiannis Aloimonos. Robots with Language: Multi-Label Visual Recognition Using NLP. IEEE International Conference on Robotics and Automation, ICRA. 2013.

Douglas Summers-stay, Ching L. Teo, Yezhou Yang, Cornelia Fermüller, Yiannis Aloimonos. Using a Minimal Action Grammar for Activity Understanding in the Real World. IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS. 2012.

Ching L. Teo, Yezhou Yang, Hal Daumé III, Cornelia Fermüller, Yiannis Aloimonos. Towards a Watson That Sees: Language-Guided Action Recognition for Robots. IEEE International Conference on Robotics and Automation, ICRA. 2012.

Ching L. Teo, Yezhou Yang, Cornelia Fermüller, Yiannis Aloimonos. Synergistic Methods for using Language in Robotics. Performance Metrics for Intelligent Systems Workshop, PerMIS. 2012. Oral

Xiaodong Yu, Cornelia Fermüller, Ching L. Teo, Yezhou Yang, Yiannis Aloimonos. Active Scene Recognition with Vision and Language. IEEE International Conference on Computer Vision, ICCV. 2011.

Yezhou Yang*, Ching L. Teo*, Hal Daumé III, Yiannis Aloimonos. Corpus-Guided Sentence Generation of Natural Images. Conference on Empirical Methods in Natural Language Processing, EMNLP. 2011. (* indicates equal contribution) Oral

Ching L. Teo, Yezhou Yang, Hal Daumé III, Cornelia Fermüller, Yiannis Aloimonos. A Corpus-Guided Framework for Robotic Visual Perception. AAAI Workshop on Language-Action Tools for Cognitive Artificial Agents. 2011.

Ching L. Teo, S. Li, L-F. Cheong, J. Sun. 3D ordinal constraints in Spatial Configuration for Robust Scene Recognition. 19th International Conference on Pattern Recognition, ICPR. 2008.

Other Publications

Talks

Current Courses and Teaching

Last Updated: January 08, 2016