Zentree Labs - Vizon Bundler: Automated Computer Vision Pipeline Designer

Vizon Bundler

Vizon Bundler is an automated event-driven contest-dependent processing pipeline designer and developer. It is a computer vision expert system. One can build vision applications for their targets defining the requirements.

Context is the one that defines the operational environment.
An event can be defined as an action in the given context
It proposes only event-based information extraction and event capsule transmission

Vizon Bundler helps in building the stack to be deployed for the purpose. Once a platform is designed can have the potential to host many contexts and events. Enhances the power of the targeted platform in a multitude.

The typical issues most companies face after they build necessary camera-based systems with necessary hardware accelerators are:

Fast adaptability of their hardware for different verticals covering different contexts and events.
Hiring and maintaining expensive teams for research and development.
Design challenges involved in using complex neural networks.
Above all we need very niche expertise in the optimization of the vision system to fit it on target.

Features

The features supported on the Vizon Bundler are:

Automated CNN model optimization
Automated Model Quantization
ALMT – Automated Lean Model Training
Resource-aware optimization and quantization
On-demand data annotations by crowdsourcing
On-demand data collection through crowdsourcing

Advantages

Fully Automated Solution (Lean Model as output)
Smaller memory footprint & Power efficient solution
Seamless porting of models on to edge devices
Minimal loss and guaranteed accuracy
Right First Time

Who Needs this? Best fit for Vizon Bundler

Private Companies/Public establishments including government involved in:

Building Surveillance solutions
Camera companies – Wants to support different applications on their devices
Vision analytics companies – Wants to achieve speed, and to fit on target and acceptable accuracy.
Embedded vision companies – Looking for Optimal models

Also, teams facing the challenges with following issues needed this solution

Not able to fit the model on their targets
Not able to control the losses in the quantization
Who wants to build hybrid models
Multiple inferences from the same input
Same inference from multiple inputs
Facing challenges adopting models for proper interfaces for HD, FHD, etc.
Companies optimizing scheduling the CNN block on their accelerators

Finally, this solution is needed for the teams:

Whoever wants to adopt the open-source models for their practical use.