After an MSc in Artificial Intelligence & Advanced Visual Computing at Ecole Polytechnique, Anh-Quan's first PhD paper "MonoScene: Monocular 3D Semantic Scene Completion" got accepted at the CVPR conference!
A little update about one of our brilliant Alumni!
Who is Anh-Quan?
Some time ago, we published an article highlighting our Alumni Anh-Quan's career. He received a MSc degree in Artificial Intelligence & Advanced Visual Computing from Ecole Polytechnique, a MSc degree in Data Science from Telecom Paris, University of Paris-Saclay and a BSc in Computer Science from University of Science and Technology of Hanoi (USTH).
Looking back at Anh-Quan previous article here.
What has become of him?
Currently a PhD student in computer vision group at RITS team, Inria Paris. His current research is focused on 3D scene understanding from 2D image(s), working under the supervision of Raoul de CHARETTE.
He submitted his first work to the CVPR conference, the top 1 conference in computer vision and machine learning. We are pleased to know that his first PhD paper "MonoScene: Monocular 3D Semantic Scene Completion" has just been accepted!
Our AI graduate published a paper on a topic he's passionate about. He wanted to share with us more information about his project and his interest. He stated:
« Through the AI-VIS master, I had been exposed to a wide range of exciting and demanding topics in AI. Among them, I am deeply interested in computer graphics, which studies how to represent and render 3D shapes digitally, and computer vision, which learns to extract information from the image. Thus, I was dreaming of aligning the two topics, making computers able to understand 3D scenes from 2D observations such as predicting the 3D shape, class, position, and orientation of the objects (e.g. car, human, tree) and the surrounding environment from images. This is a challenging and long-lasting research problem despite being trivial for humans. Therefore, I decided to pursue a Ph.D. in 3D computer vision with the goal of developing AI methods that can reconstruct the complete 3D scene from a single image. Since these methods only require cameras, it would pave way for a wide range of applications in VR, AR, and Robotic to be shipped in smartphones, drones, and cars which traditionally required less compact, more expensive and more intrusive 3D sensors. »
