Khoa Le Tan Dang Email & Phone Number

Singapore

Object DetectionDesigned annotation workflow for data collection, conducted research and implemented ML models; successfully deployed detection models to serve multiple Visenze's solutions (visual search, tagging, recommendation).Projects: general object detection, real-time on-device detection.Vector Search * Conducted research on state-of-the-art similarity search systems. * Worked with API team to improve the existing search system efficiency by introducing new compression methods and low-level optimization. * Developed a cross-modal retrieval for product de-duplication and image matching. * Implemented and deployed a new rerank logic for the search system.Augmented Reality: Played many roles in the development including designing data collection tools, training and deploying ML models for mobile devices, designing system logic.Notable projects are: * Annotation tool for 3D box labeling. * Real-time virtual shoe try-on app on iOS. * Successfully delivered SDKs for 3D objection detection, 2D tracking, hand gesture recognition.

Singapore

I have worked for two years as a major developer of the computer vision project funded by Ministry of Defense and conducted research on deep learning image retrieval.Projects: Urban-Area Scene-Based Localization: - Developed the localization system on Android mobile devices with high accuracy: 85% rate recall. - Translated the Matlab simulation to Java and C++ code. - Optimized the system in order to gain low latency and reduce memory consumption, i.e, it can localize a scene by Structure-from-Motion < 3s per 3D model while obtaining an on-device database of 15000 images.Tools: Matlab, C++, OpenCV, Eigen, Boost, CUDA, OpenGL, Android NDK, Java, VisualSFM.Research Interests:- Image content-based retrieval.- Image Hashing.- Deep Learning.

Singapore

Project: “Urban Building Image Retrieval” (sponsored by Ministry of Defense).Design and implement Android framework for Image Retrieval System.

Saskatchewan, Canada

Lymphocytes counting system: - Developed a new method for counting lymphocytes. - Proposed a new image feature learned from color channels and contours and adapted a fusion classifier from SVM for classification. - The system delivers high accuracy over 90% with low execution time (<1s on typical computers).Tools: Python, numpy/scikit-image, sklearn, OpenCV, Qt.

Nomi, Japan

Project: Image segmentation in vision-based hand gesture recognition.I conducted the literature survey about RGBD segmentation for hand gestures via Kinect and LeapMotion devices.Programming Language: C++, C#, Python.Tools: Kinect SDK, numpy, OpenCV, OpenNI, LeapMotion.

Ho Chi Minh

* Developer in Online library project.* Manage and develop center's websites.Tools: PHP/Codeigniter, Javascript, jQuery, MySQL.

Master'S Degree, Artificial Intelligence

Bachelor’S Degree, Computer Science, Honors Program