TAPPAS
Hailo’s TAPPAS (Template APPlications And Solutions) is specially designed to simplify high-performance edge applications development and deployment, based on the industry-leading Hailo-8™ AI processor, and allowing customers to accelerate time to market by reducing development time and effort. It provides an easy-to-use GStreamer-based template for application development, helping users to quickly deploy their own networks on the Hailo-8™.
TAPPAS is an infrastructure containing a suite of high-performance pre-trained template AI tasks and applications with pipeline elements, built on top of state-of-the-art deep neural networks, demonstrating Hailo-8™ best-in-class throughput and power efficiency.
The open applications infrastructure offers different operational modes, bring-your-own-model, and DIY support functionalities, which greatly help achieve development flexibility for a variety of categories and industries.

TAPPAS out-of-the-box versatile apps include:




Learn More
Classification
Ready-made ResNet-50-based application, trained on ImageNet




Learn More
Depth Estimation
A ResNet18-UNET-based application, rudimentary in intelligent ADAS cameras




Learn More
Lane Detection
Out-of-the-box PolyLaneNet-based application trained on Tusimple for basic assisted driving applications




Learn More
Semantic Segmentation
Purpose-built and pre-trained model, optimized for high-resolution ,real-time applications like ADAS front-facing cameras




Learn More
Multiple Object Tracking
An advanced versatile model that detects and tracks multiple simultaneously-moving people




Learn More
Facial Detection & Recognition
Apps for face recognition and detection app for security and authentication applications




Learn More
Pose Estimation
CenterPose-based app for a variety of safety, security and operational optimization applications across verticals




Learn More
Object Detection
Detection apps with several available neural networks, delivering unique functionalities




Learn More
Instance Segmentation
A YOLACT-based model that identifies, outlines and colors different objects and persons