Computer Vision Company Chooch AI Makes Breakthroughs in Edge AI in Speed, Accuracy and Scale

Speeds of 20 millisecond inference time, 8 models and 8,000 classes on even the smallest NVIDIA Jetson Nano edge devices - ultra-low latency application Chooch brings powerful Edge AI to Intelligent Video Analytics for the Enterprise


SAN FRANCISCO, Sept. 17, 2020 (GLOBE NEWSWIRE) -- Chooch AI, a horizontal computer vision AI platform, today published the latest video presentation of an expert panel with Convergint Technologies and NVIDIA. Chooch AI is a member of the NVIDIA Inception program. For the full transcript and video, please visit Edge AI Breakthroughs.

The flexible AI training dashboard ingests any type of visual data and generates up to 1,000 annotated images per minute with video annotation, enabling base models to be built in hours. Chooch AI has been deployed at scale for 18 enterprise clients and the US Government, with applications in aviation, media, geospatial/GIS, healthcare, security, retail and industrial IoT.

Breakthroughs include:

  1. Speed: Inference speeds as fast as 20 milliseconds allow for real-time AI.
  2. Remote edge device updates: AI models are deployable at scale to edge devices from a single dashboard with configurable reports and alerts sent via text, email or API.
  3. Fully deployable AI models: Pre-trained models are now available which do not require customer data, for workplace safety, fire detection, crowd analysis, and more.
  4. 8000+ AI classes & 8 models on NVIDIA Jetson Nano: Accuracy at higher than 90% with ensemble models and robust sets of classes.
  5. Inferencing on Multiple Camera Streams: models can inference on up to five camera streams simultaneously with the NVIDIA Jetson AGX Xavier.
  6. Configurable real time alerts and reports: Chooch Edge AI Dashboard manages models, devices, cameras and settings for alerts and reports.
  7. Rapid flexible training: The AI training dashboard ingests any type of visual data for video annotation and generates up to 1,000 labeled images per minute, allowing base models to be built in hours.

Unlike single-purpose computer vision systems, Chooch Al can be quickly trained with new data without coding skills. Fast model development and deployment mean that Chooch AI can be quickly trained to perform inferencing for a wide variety of objects, actions, processes and states. Organizations have the choice to deploy the Al models in the cloud or onto edge devices, where the system can be configured and integrated with business intelligence dashboards.

Datasheets, videos and papers available at Computer Vision Resources.

To see Chooch AI in action, please visit Edge AI Breakthroughs or get an AI Demo

And follow us on Twitter at @chooch_AI

About Chooch AI
Chooch AI is a powerful visual Al platform that replicates human visual tasks and processes by generating fast, accurate computer vision services across a wide variety of applications. Chooch proprietary technology has been deployed at scale for industries including aviation, media, geospatial, healthcare, security, industrial lot, and the US government. Applications range widely from tracking surgical procedures, counting store employees, touchless check-in, security for remote oil & gas sites, facial authentication, intelligent video analytics, and image quality control. For more information or a technology demonstration, please visit www.chooch.ai or contact press@chooch.ai

 

Coordonnées