This South Korea-based AI startup, Nota AI, is revolutionizing the AI space with its proprietary hardware-aware AI optimization platform to automate the development process of lightweight AI models

Pitch your startup story at [email protected] 

Please don't forget to join our ML Subreddit

Need help in creating ML Research content for your lab/startup? Talk to us at [email protected]

We have a plethora of AI models being developed by numerous businesses. But these require a lot of time and resources. Because high-performance AI models are rapidly rising in size, and most AI models are too large for edge devices to manage, AI model optimization tools are now more important than ever.

Nota AI is bridging this gap with its technology for optimizing AI models and edge device AI solutions. It provides an end-to-end platform to develop near production-level edge AI models, reducing model development time from months to weeks.

NetsPresso, Nota’s flagship product, is a proprietary hardware-aware AI optimization platform for lightweight AI model development. NetsPresso, currently in the beta testing stage, significantly decreases the time and resources needed to create an AI model while optimizing it for the target device. Netspresso consists of the following tools to make the hardware-aware AI model optimization process more efficient.

• Nota is releasing the beta version of ‘Model Search’ on April 20th. This is the hardware-aware autoML module that can automatically search optimized models for a target device with just the input of a dataset.
• The beta version of the ‘Compression Toolkit‘ module was released in late 2021 and is open to beta testers. The module is a ready-to-use toolkit for AI developers and deep learning engineers looking to compress their AI models. With this toolkit, one can compress the AI models in just a few clicks. 
• They are also preparing to launch their next module, ‘Model Packaging,’ in July. This module is a ready-to-deploy package to accelerate and deploy the model on hardware immediately.

Nota’s services allow users to customize AI model compression using user-selected parameters, such as optimization for specific devices based on Nvidia, Raspberry Pi, and other suppliers. In comparison to the unoptimized model, Nota claims that inference performance can be boosted eight-fold while consuming less power.

Nota’s AI optimization technology enables a diverse range of edge AI solutions including intelligent transportation system and low-powered driver monitoring system. Real-time lightweight ITS solutions based on the company’s AI technology improve traffic flow, saving city traffic-congestion-related costs and reducing the time spent by drivers stuck in traffic. As for DMS, the solution detects unregistered drivers, driver drowsiness, cell phone usage, and driver distractions like smoking on low-end devices based on our lightweight AI technology.

The tech startup raised $14.7M in a Series B funding round and additional funding from Kakao Investment recently. The company plans to use this recent investment funding for enhancing its platform as well as expanding into new markets across Europe and North America. Nota AI will be exhibiting as a gold sponsor at the 2022 Embedded Vision Summit at booth 418 and AutoML Conference 2022, where they will be showcasing the aforementioned NetsPresso platform and its sub-modules.

References: