Frederic Lefebvre

Background

About Frederic Lefebvre

During my PhD period, I pointed out an innovative solution for content identification and indexing based on image hashing for multimedia content. Once I joined Technicolor (Thomson) in 2005, I launched the video fingerprint (large scale) activity inside Technicolor and took the leadership of the fingerprinting team in Security Lab and was involved in the video watermarking solution (transfer to ContentArmor). I proposed also selective encryption for multimedia content (h264 and Then, I discovered new areas in computer vision and color science. I contributed to color re mastering of legacy content, color/look transfer. I was working on deep learning (Neural Network) by exploring and deploying (libtorch c++, OpenCV DNN) new algorithm for video Rotoscoping techniques in order to help the artists to track (optical flow) and segment (Detectron2/Pointrend) the object in real scene for 2K, 4K and 6K contents. Now, I am working on Neural Network (compressAI) and conventional video compression tools for video compression standards VVC (JVET). I am the author of more than 20 research articles and granted more than 10 patents. & hl=fr & oi=ao Specialties: Video Fingerprinting, Image Video processing, Watermarking, Selective encryption, color science and computer vision, image and video segmentation, superpixel, optical flow, deep learning, compression