Tag: #pose-estimation6 results
A HEMA practitioner is creating an open-source, high-frame-rate dataset to address computer vision challenges in tracking fast-moving, occluded objects like swords. The project aims to improve pose estimation and trajectory prediction for complex, high-speed physical movements.
BioCoach is an AI prototype that monitors exercise form via camera. It reconstructs the user's skeleton in 3D to provide real-time corrections for joint angles.
Google Cloud and Google DeepMind jointly developed an AI system to 3D analyze complex aerial maneuvers in skiing and snowboarding. The system estimates athletes' postures and has been provided to the US team for the Milan-Cortina Winter Olympics. It also holds potential for rehabilitation and performance enhancement.
This paper introduces a periodic adaptation framework for pose-based, unsupervised shoplifting detection deployable on IoT edge devices in retail. It releases RetailS, a large-scale real-world dataset from multi-day, multi-camera store surveillance. The approach outperforms offline baselines in 91.6% of cases with training under 30 minutes on edge hardware.
WiFi DensePose, an open-source project with 17,000 GitHub stars, uses standard WiFi signals to detect human poses through walls and monitor breathing and heart rate. It eliminates the need for cameras or wearable devices. This innovation unveils new potentials in intelligent perception.
Nex Playground, an AI-powered motion-sensing game console from ex-Apple engineer David Lee, captured 14% US market share on Black Friday 2025, surpassing Xbox. Leveraging AI pose estimation from its basketball app roots, it targets parents with kid-friendly IPs like Bluey and a hardware+subscription model. 2025 revenue exceeds $150M, with 2026 goals at $500M.