Watch the pitch videos from the 2023 finalists!
These start-ups are bringing new innovations to the computer vision and visual AI market.
Judges' Award Winner
ProHawk Technology Group
ProHawk Technology Group is a computer vision restoration solutions provider that offers cutting-edge solutions to businesses across various industries. Our mission is to help our clients optimize their operations and increase their bottom line by leveraging computer vision restoration technologies.
ProHawk Technology Group, Inc. is a computer vision restoration solutions provider that offers cutting-edge solutions to businesses across various industries. Our mission is to help our clients optimize their operations and increase their bottom line by leveraging computer vision restoration technologies. ProHawk’s technology has diverse applications across a multitude of industries, spanning Safety and Security, Healthcare, Manufacturing, Retail, Government, and Transportation. Our partnerships validated designs and market alignment with industry leaders, such as NVIDIA Metropolis, NVIDIA Clara Holoscan, and Dell Technologies Computer Vision Safety Security and Healthcare teams, have positioned us as pioneers in the field. ProHawk Technology Group is a leading computer vision solutions provider offering cutting-edge solutions that assist businesses in streamlining their operations and enhancing their profitability. ProHawk’s technology is leading the way in computer vision restoration and contribute to a safer, more efficient, and more innovative world.
Audience Choice Award Winner
GMAC Intelligence
GMAC is building a drive-through automation solution for quick-service restaurants (QSRs). We are solving the labor shortage and churn problem for the US QSR industry. Our solution is QSRBot 247 for drive-through automation, which identifies loyal customers, takes their food orders accurately, handles their payments and delivers the food quickly. This solution is enabled by on-device implementation of vision AI, ASR, TTS and NLP algorithms.
GMAC is building Drive-thru Automation Solution for Quick Serve Restaurants. We are solving the labor shortage and churn problem for the US QSR industry. Our solution is QSRBot 247 for drive-thru automation, which identifies loyal customers, takes their food orders accurately, handles their payments and delivers the food quickly. This solution is enabled by on-device implementation of vision-AI, ASR, TTS and NLP algorithms. We leverage off-the-shelf Edge-AI hardware platforms (e.g., Qualcomm) which have smartphone unit economics to enable the lowest capex/opex solution for this market. In addition, we have integrated with POS billing systems and ChatGPT as well to provide an E2E, robust and plug-n-play solution. GMAC is building a software solution enabled on a 3rd party Edge-AI platform with a SaaS business model. This is a 24 BUSD/yr market worldwide (TAM).We have enabled on-device Vision for on-device Chatbots (ASR and TTS and NLP) and ChatGPT. Our long-term goal is to enable on-device LLM as well.
Finalists
Lemur Imaging will be showcasing a real-time, interactive computer vision demo of their ultracompact Line Memory Reduction (LMR) hardware IP. The demo will show how LMR can compress activation data by 60-75%, saving both silicon area and DRAM bandwidth, with no compromise on detection quality.
Lemur Imaging, a UK based start up, founded by industry veterans, Noman Hashim and Dr. Slava Chesnokov (ex ARM/Apical), to address the growing problem of reducing system cost and power consumption whilst increasing Computer Vision detection accuracy. The first product, available now, is LMR™ (Line Memory Reduction) a novel ultra compact (sub 20k gates),low latency (sub 10 clocks), perceptually lossless compression core, which allows computer vision engineers to improve the accuracy of their models whilst reducing SRAM by 2x and DRAM bandwidth/power consumption by 3-4x. LMR™also introduces the concept of LMR5 data type – allowing companies to take FP16 images and compress down to 5bits, thus reducing the activation data stored in SRAM by 70% and/or reducing DRAM bandwidth by 3x. LMR5 data does not require for any model retraining yet delivers significant power saving and increased detection accuracy when compared to FP8/INT8 based solutions.
Optimizing Mind has developed a novel technology called Transfer Learning Squared that performs transfer learning training much more efficiently, using 1% of data samples. This technology overcomes the problem known as catastrophic forgetting and is an easy way to reduce development time and costs.
While AI is powerful, it can be challenging to train especially for highly dynamic edge scenarios. For robust applications, developers must obtain good data and qualified, seasoned AI experts for training this data. Optimizing Mind has developed a novel technology called Transfer Learning Squared that performs transfer learning training much more efficiently using 1% of data samples. Moreover this technology overcomes the problem, known since the early days of AI as catastrophic forgetting. This means that not only less samples are needed, but the order of the samples does not matter allowing the AI to be more easily debugged, edited and updated. Transfer Learning Squared is offered as a SaaS and is seamlessly adoptable because it does not require a change to current workflows. It is an easy way to reduce development time and costs.
Polly is an AI-enhanced assistive platform that helps people with neurological disorders such as MS, CP, Alzheimer’s and Parkinson’s. Polly has an AI/ML platform that gives real-time voice (communication), cognitive assistance (control) and telecare (caregiving) to people with advanced neurological conditions.
Polly helps people with neurological disorders such as MS, CP, Alzheimer’s and Parkinson’s, by AI/ML platform giving a real-time voice, cognitive assistance and telecare to people with neurological conditions. Polly provides three values (3 C’s): 1-Communication: Real-time voice and facetime based on the contextual surroundings. 2-Control: cognitive assistance, computer vision, controlling IoT Devices, memory, and behavior assistant. 3-Caregiving: Assisting caregivers with remote patient monitoring and alarm system.
