Tag: Computer Vision

Researchers have built InterPet4D, the first large-scale multimodal 4D dataset of human–dog interactions, and used it to train an AI ...

Researchers have built an AI framework that combines live video of floc formation, real-time turbidity sensing and physics-based equations to ...

Researchers in Vietnam have developed WGPose-LL, a deep learning method that uses paired well-lit images to guide human pose estimation ...

Researchers at Shiv Nadar University have developed a dual-path AI framework that combines a residual gated autoencoder with patchwise nearest-neighbor ...

Researchers have developed a lightweight frequency-based AI detector that achieves state-of-the-art accuracy and real-time speed on cluttered traffic scenes, running ...

A new lightweight deep learning framework called LCASP-MMANet achieves state-of-the-art sign language recognition accuracy while remaining efficient enough for real-world, ...

Researchers used a consumer tablet app to capture facial blendshape features that objectively quantify Parkinson's hypomimia, revealing asymmetry and interaction ...

Researchers at Dalian University of Technology have developed PD-Dehaze, a progressive diffusion framework that restores haze-degraded photographs with stable coarse-to-fine ...

Researchers at the University of Essex and British Telecom have built an AI video auditing system that combines fuzzy logic, ...

A controlled benchmark from El Salvador found that the lightweight hybrid MobileViT-v2-100 model reached 96.45 percent accuracy in classifying sugarcane ...

Researchers have combined an Enhanced-Value attention scheme within a Swin Transformer with super-resolution-guided training to improve detection of tiny objects ...

Researchers at MIT's Senseable City Lab and international collaborators have developed a deep learning framework that converts street camera video ...

A hybrid quantum-classical detector called Q-SODA uses an eight-qubit entangling circuit to achieve record precision on small object detection while ...

Researchers at Harbin Normal University have developed WFGformer, a dual-branch transformer network that uses Haar wavelet frequency decomposition to restore ...

Researchers have unveiled a triple-stream deep hashing framework that compresses multi-label images into compact binary codes with higher accuracy and ...

A new multimodal AI framework fuses RGB images, segmentation masks, depth maps, and text prompts to estimate robotic gripper pressure ...

Researchers in Mumbai have developed a wedge-based shadow analysis technique that detects image forgeries by testing whether the shadows cast ...

A new survey compares marker-based, markerless and thermal imaging approaches to human pose estimation for detecting gait changes linked to ...

Researchers in Chongqing have built a lightweight camera-only 3D detection framework that adapts its processing to scene complexity, fuses information ...

A foundation model-based AI system called WeedNet identifies 1,593 weed species with 91.02 percent accuracy and reaches 97.38 percent accuracy ...