Abstract: With the rapid development of intelligent transportation systems and growing emphasis on driver safety, real-time detection of driver drowsiness has become a critical area of research. This ...
Abstract: Recent advances in artificial intelligence (AI) models, such as large language models and diffusion models, have shown significant potential in semantic communication by reconstructing ...
Synthesizing realistic audio, images, and videos using algorithms has always been essential in Signal Processing, Computer Graphics, and Computer Vision. When using pre-artificial intelligence (AI) ...
NVIDIA diffusion language model Nemotron TwoTower achieves 2.42x LLM inference throughput without a full retraining run, ...
Deploying DFlash block diffusion on NVIDIA hardware accelerates autoregressive LLMs during latency-sensitive inference.
Outperforms advanced methods in terms of rate-distortion-perception performance. Delivers exceptional encoding efficiency for 35.8 FPS@1080P Maintains competitive decoding speed compared to existing ...
Official implementation of Whisfusion - the first Diffusion Transformer ASR framework that fuses a Whisper encoder with a diffusion decoder for faster, non-autoregressive transcription.
The hierarchical diffusion model requires effective conditioning on both spatial perception and proprioceptive states. A naïve concatenation of conditioning variables with action sequences is ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results