PINN Method Breaks Through Bottleneck in Nanobeam Bending Analysis
A new study proposes the DFL-TFC framework based on physics-informed neural networks for analyzing the bending behavior …
📂 Research
Latest articles in Research
OAMVOS: Cracking the Occlusion Challenge in Video Object Segmentation
Addressing the fragility of SAM-based dense trackers in scenarios involving prolonged occlusion and fast motion, OAMVOS …
From Skeletons to Pixels: A New Few-Shot Method for Precise Event Spotting
A latest arXiv paper proposes two complementary distillation strategies — representation distillation and prediction dis…
AgentRVOS: Agent-Based Video Object Segmentation Solution Wins Third Place at PVUW Challenge
In the MeViS text track of the 5th PVUW Challenge, the AgentRVOS solution built around Sa2VA introduced an explicit agen…
New Paradigm in Neural Network Optimization: Decoupled Training and Fine-Tuning Strategies
A latest arXiv paper proposes a decoupled optimization technique that redesigns neural network optimizers to address the…
Train Visual AI Models in Your Browser, Deploy to a $15 Microcontroller
A new arXiv paper introduces the webmcu-vision-web tool, enabling users to complete end-to-end TinyML vision model train…
ParkingScenes: Filling the Gap in Autonomous Parking Datasets
A research team has released ParkingScenes, a structured dataset specifically designed for end-to-end autonomous parking…
Breakthrough in Cross-Modal Representation Learning: Fusing Imaging and Transcriptomics to Accelerate Drug Discovery
A latest arXiv paper proposes an intervention-aware multi-scale representation learning framework that deeply integrates…
Vision Language Models Completely Fail Dynamic Gauge Reading Tests
A latest arXiv paper reveals that current mainstream vision language models (VLMs) suffer from severe reading accuracy d…
MetaEarth3D: Global-Scale 3D Generation Technology That Breaks Spatial Limits
A latest arXiv paper introduces the MetaEarth3D framework, which employs a spatially scalable generative modeling approa…
New Research: Boosting 3D Human Pose Estimation with 2D Pretraining
A latest arXiv paper proposes using 2D pretraining strategies to improve 3D human pose estimation, breaking the traditio…
Study Reveals Link Between Infrasound Exposure and Negative Emotional Responses
A new study shows a significant association between infrasound exposure and aversive responding as well as negative appr…