Records found: 6
Explore a detailed tutorial on building an end-to-end object tracking and analytics system using Roboflow Supervision, combining detection, tracking, and zone-based analytics with real-time annotations.
A recent EPFL study benchmarks multimodal foundation models like GPT-4o on core vision tasks, revealing strengths in semantic understanding but highlighting gaps compared to specialized vision models.
Buildots raised $45 million to enhance its AI-driven construction management platform, featuring real-time tracking and an AI assistant, aiming to revolutionize efficiency and cost control in large-scale construction projects.
Tencent introduces PrimitiveAnything, a novel AI framework that reconstructs 3D shapes by sequentially generating geometric primitives, enhancing semantic understanding and generalization across diverse objects.
MirrorVerse and MirrorFusion 2.0 introduce innovative datasets and training techniques that significantly enhance diffusion models' capacity to generate accurate and photorealistic mirror reflections.
NVIDIA introduces Describe Anything 3B, a multimodal large language model that excels in detailed, region-specific captioning for images and videos, outperforming existing models on multiple benchmarks.