1st1 Technologies’ disruptive algorithm-based image matching technology, DejaVuAI®, is available to users via Photon. It can be integrated into the NN environment to reduce in-pixel data complexity and improve the accuracy and throughput of image recognition. This novel approach to image recognition, uses their proprietary methods to index and process images that significantly reduces time, cost, and effort.
1st1’s algorithm, DejaVuAI®, relies on CPU, RAM, and cache—no GPU or Tensor cores are needed. It can be integrated into any environment to process images immediately upon collection; and was designed for a multitude of deployment types, including near-sensor processing. As a result, regardless of frame rate or format researchers mass amounts of image data is searchable without the requirements of metadata tagging, object classification, and inference processing. The Photon User Interface (UI), powered by DejaVuAI®, gives researchers a powerful tool that results in exponentially faster and more accurate image recognition than other machine learning (ML) and artificial intelligence (AI) methods and technologies available on the market today.
The DejaVuAI® algorithm reduces hardware costs, power consumption, and human effort required to perform image recognition. Moreover, dewarping and advanced image types are integrated and supported, including floating point, stereoscopic, multi-view, and 3D imagers. 1st1’s approach to AI processing is already highly accurate and requires less compute power.
Initial testing used an image set that contained a 1M image FLICKR sample dataset (~120GB) including a 10k (3.66GB) image set of famous artwork. The results from the ML image search engines collectively had 0% accuracy. However, using Photon, the results were 100% accurate with image registration, confidence score, rotation, and scale.
While Low Latency is subjective, the test search using Photon took 2 minutes to search through +1M high def images with arbitrary rotation on a $1,000 personal laptop with 100% accuracy. No GPU or Tensor core was needed or used.
We believe the DejaVuAI® algorithm has the potential to greatly impact and enhance searching effectiveness anywhere still or video images. For more information, contact us now.
