Max Robotics — Global Service Robot Directory + US Certification

About

Junyu Dong is a prominent computer vision and machine learning researcher whose work spans underwater imaging, photometric stereo, object detection, and depth estimation. He is perhaps best known for his pioneering contributions to underwater computer vision, where his development of the Invert Multi-Class Adaboost framework combined with deep learning addressed critical challenges in detecting small, blurry, and low-contrast underwater objects — work that has garnered over 118 citations and established him as a leading voice in marine AI applications. His research extends to underwater surface normal reconstruction via transformer-based photometric stereo techniques and the adaptation of large foundation models, such as SAM, for underwater object segmentation, reflecting his sustained commitment to advancing ocean robotics and 3D data acquisition. Beyond underwater imaging, Dong has made notable contributions to facial expression recognition using cascade regression-based frontalization, monocular depth estimation through two-stage deep regression, and transparent material detection — demonstrating impressive breadth across applied vision tasks. His earlier work on efficient image rotation for Chinese OCR (2006) highlights a career-long dedication to practical, real-world solutions. With a consistently cited body of work across multiple domains, Dong's research has meaningfully shaped how deep learning handles complex, real-world visual environments, making him an influential figure for students and researchers navigating computer vision today.