Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Waseda University Tokyo
3
0
3
Accelerating learned image compression by over 13 times without retraining could revolutionize real-time applications in computer vision.
Achieving 122 fps with 10-bit 4:2:2 encoding on commercial GPUs could revolutionize real-time volumetric video streaming capabilities.
Achieving superior image compression performance by navigating the full rate-distortion-perception surface without extra rate costs could revolutionize how we approach neural image codecs.