Megh Computing Uses CAST H.264 IP Subsystem in Video Analytics Solution
01/08/2020 - 15:08
Low-latency Video Decoding IP Subsystem Meets Megh's Requirements for Video Analytics System Using AI/Deep Learning to Prevent Retail Inventory Loss
Megh Computing
Megh Computing, established in 2017, is based in Portland, OR with offices in Bangalore, India.
Megh Computing’s mission is to enable the 3rd wave of computing in data centers with the deployment of scale out FPGA (Field Programmable Gate Arrays) based accelerators.
The company’s solution provides a scalable, efficient platform for Real Time Analytics using FPGA-as-a-Service delivering high performance with low latencies for the edge, public or private cloud. For more information, visit www.megh.com.
The Challenge
Megh Computing had both a technical and business challenge. The technical challenge was very tight timing and performance specs.
As a start-up, the business challenge was to come up with a flexible licensing model that met both Megh Computing’s current and future economic requirements.
The Solution
Megh Computing uses the CAST Video Over IP subsystem that integrates a CAST H.264 decoder with the CAST RTP-to-H264 core for a low latency streaming solution.
This IP solution is used to decode video streams into image frames for segmentation and classification using deep learning on the streaming video. The application of this product is to use AI/Deep Learning to prevent retail inventory loss and to track inventory items in warehouses. Read Megh Computing’s blog on Using AI/Deep Learning to prevent inventory loss to get a more detailed description of their solution.
Megh Computing's Requirements for their video analytics pipeline.
What attracted Megh Computing to CAST was CAST’s technology and domain expertise, their support model and willingness to work with Megh Computing to meet Megh Computing’s stringent timing requirements and project schedule with CAST’s integrated H.264 subsystem. CAST worked with Megh Computing to provide a licensing model that met the needs of both a start-up and projected future growth. Megh Computing has recently demonstrated its Video Analytics Solution on the Intel Platform.
Key Challenges
Needed a partner with strong domain expertise.
Needed a partner who was willing to work with Megh Computing to meet their stringent timing requirements and project schedule.
Met and exceeded Megh Computing’s timing requirements and performance requirements.
CAST provided a flexible licensing model.
Learn more about the challenges Megh's engineers faced and their solutions in this talk from the Spark+AI Summit 2019: Accelerating Real Time Video Analytics on a Heterogenous CPU + FPGA Platform
by Bhoomika Sharma
A negligible delay from video capture to display—the “glass-to-glass” latency—is critical for many video streaming applications, from advanced driver assistant systems (ADAS) and unmanned vehicles to video conferencing and broadcasting.
In the video world, latency is the amount of time between the instant a frame is captured and the instant that frame is displayed. Low latency is a design goal for any system where there is real-time interaction with the video content, such as video...
