class in: Science and technology, transportation
Subjects : Awards and Rewards, New Products and Services, MRR
HANGZHOU, China, Jan. 18, 2023 /PRNewswire/ — As the core algorithms of artificial intelligence, visual object detection and tracking is widely used in home surveillance scenarios. Recently, IMOU, the Chinese home automation brand, won the top spots in KITTI evaluations for 2D object detection (socket) and multi-object tracking (socket and car). It achieves 57.15% High Order Tracking (HOTA) accuracy for pedestrian multi-object tracking, 82.08% (HOTA) for auto multi-object tracking, and 82.77% (moderate) for 2D pedestrian object detection.
Generic object detection and tracking are fundamental and difficult tasks in computer vision. For 2D object detection, IMOU provides an algorithm framework for structural reparameterization and multimodal fusion, enabling significant performance gains in occlusion and small objects. This framework has been fully applied to IMOU smart cameras.
For tracking multiple objects, IMOU optimizes object detection, re-identification, object correlation and other modules based on detection tracking framework to improve overall performance. In particular, the recognition ability based on the structural reconfiguration is improved. Furthermore, we achieve a significant performance gain in distortion and occlusion cases by providing a dynamic model update strategy. In addition, the graphics network is used to model target tracking timing sequences to strengthen multi-function fusion and improve multi-object tracking stability. With the help of the proposed object tracking algorithm, IMOU cameras can continuously track targets of interest and record the trajectory of movement. If abnormal behavior is detected, our users will receive an alert.
Take Cruiser 2, an AI outdoor camera from IMOU. Local AI algorithms are equipped with powerful AI chips for efficient data analysis and operation. Newly updated AI capabilities can accurately detect pedestrians and vehicles, continuously track points of interest, and record the path of movement. In addition, the monitored area can also be intelligently divided on the APP, increasing the flexibility of intelligent monitoring.
The KITTI Object Recognition Benchmark is currently one of the largest data sets for evaluating computer vision. It was founded jointly by the Karlsruhe Institute of Technology in Germany and the Toyota Research Institute in the United States. KITTI is used to assess stereo vision, optical flow, scene flow, visual odometer, object detection, target tracking, route detection, and semantic and instance segmentation. As a complete home automation solution provider, IMOU has been working in the field of AI for years and keeps making groundbreaking discoveries. On the one hand, our AI technology is constantly improving household products in terms of high intelligence and humanity. On the other hand, we are trying to change the situation where household products are isolated from each other. In the future, more and more IMOU products will be connected, which will make it easier to build a complete home automation ecosystem, and create a simpler, smarter and safer life for every user.
Photo – https://www.spamchronicles.com/wp-content/uploads/2023/01/IMOU-Wins-First-Places-in-KITTI-Reviews-for-2D-Object.jpg
Photo – https://www.spamchronicles.com/wp-content/uploads/2023/01/1674026798_803_IMOU-Wins-First-Places-in-KITTI-Reviews-for-2D-Object.jpg
Photo – https://www.spamchronicles.com/wp-content/uploads/2023/01/1674026800_614_IMOU-Wins-First-Places-in-KITTI-Reviews-for-2D-Object.jpg
You might also be interested in these press releases
|
Press release sent on January 18, 2023 01:36 and distributed by: