When building real-time vision applications for mobile, embedded, or edge devices, developers often face a tough trade-off: accuracy versus efficiency.…
DynamicViT: Slash Vision Transformer Compute by 30% Without Sacrificing Accuracy 641
Vision Transformers (ViTs) have revolutionized computer vision, but their computational demands remain a major barrier for real-world deployment—especially on edge…
EGVSR: Real-Time 4K Video Super-Resolution with High Visual Quality and Edge Deployment Ready 945
Video super-resolution (VSR) has long promised to breathe new life into low-quality video content—enhancing resolution, restoring detail, and eliminating the…
Mengzi: Lightweight, High-Performance Chinese Pre-Trained Models for Efficient NLP Deployment 540
In recent years, pre-trained language models (PLMs) have revolutionized natural language processing (NLP), delivering state-of-the-art results across a wide spectrum…
MambaIR: High-Quality Image Restoration with Efficient State-Space Models 966
Image restoration—recovering clean, high-resolution images from degraded inputs—is a foundational task in computer vision with applications ranging from smartphone photography…
NVIDIA FLARE: Build Privacy-Preserving AI Across Organizations—Without Moving Data 847
In today’s data-driven world, organizations often face a fundamental dilemma: they want to build powerful, generalizable AI models, but their…
Fast-BEV: High-Speed, High-Accuracy Bird’s-Eye View Perception for Real-World Autonomous Driving Systems 754
Fast-BEV emerges as a compelling solution to a longstanding challenge in autonomous vehicle (AV) perception: achieving both high inference speed…
CodeI/O: Boost LLM Reasoning Across Domains by Learning from Code Input-Output Patterns 563
Large language models (LLMs) have demonstrated impressive capabilities in code generation and narrow reasoning tasks like mathematics. Yet, when it…
SEED-Voken: Scalable, High-Fidelity Visual Tokenization for Autoregressive Image and Video Generation 984
SEED-Voken is an open-source toolkit developed by Tencent ARC that delivers state-of-the-art visual tokenizers tailored for autoregressive visual generation. Built…
TSMixer: A Lightweight, High-Performance Alternative to Transformers for Multivariate Time Series Forecasting 738
Time series forecasting is essential across industries—from predicting energy demand and stock trends to managing supply chains and monitoring IoT…