Sound Analytics Deployments at Global Level
Japan: Deployed Equilips for 24x7 monitoring of multiple sawing machines and factory level visualization – together with Konica Minolta
Working with a large Japanese customer for Ultrasonic Analysis to detect structural health (e.g. poles)
A global automotive supplier: Quality monitoring of ultrasonic welding
Sharada Industries, Pune: Welding process monitoring
In discussion with a welding machine OEM
Multiple large Indian companies in pipeline
Other applications: Milling machines, air/gas leakage, rotating equipment (motors, pumps etc.)
Equilips 4.0 Deployment in Japan with Konica Minolta
Deployed Equilips for 24x7 monitoring of Sawing Machines at a factory in Japan
Our Solution Displayed at Japan AI Summit (AISUM) Tokyo, April 2019, by Konica Minolta
Sound as primary signal, Edge devices installed
Factory level monitoring and visualization using sound analytics
WebApp developed for real-time monitoring and data visualization
Video Analytics Demo Identification of Objects in Video
Identification of machine movements and status
Identification of workers and their productivity
Identification of safety practices
Identification of defects
Scene and event identification
This can be done at high speed: More than 10fps on live video!
Video Analytics for Automotive Part, using 3 camera
A 3-camera system deployed at Sharada Industries, Pune
Defects in parts detected in real-time, as the parts move on conveyor belts
Three cameras help identify detects in any of the three sides
An automated conveyor that separates OK vs Not OK parts based on AI predictions
Video Analytics for Bottling Plat
For a large Japanese beverages company, developed video analytics solutions for their plant in Thailand.
Real-time video analytics (at 10 FPS) on edge for detecting 5 types of errors in bottling operations (without stopping the operations)
100% accuracy (100% recall and 100% precision) on all 5 error events!
Combination of Computer Vision, Deep Learning with Localization, and One-shot learning for good recall even with limited data!