Analyze 2000 params of a target
Analyzing 2000 parameters of every given person our system provides the most robust solution on the market to track and find suspects or missing people. Our convolutional neural networks dynamically assess a holistic set of features for every individual in each captured frame.
Traces AI carefully extracts all information from every frame where a person is detected to accurately identify an individual by assessing all visible parameters including color, type, and details of clothing, body shape, hairstyle - even items in their possession.
Auto Learning in the Field
Building a single general algorithm is not enough to fit various customers needs because usually, the model fails to generalize well to other domains. In order to deliver the best in the market accuracy for each client, our engineers developed a high-end auto retraining pipeline that learns in the field and robust to any scenario. It can adapt to different use cases whether cameras are placed indoors or outdoors, with different weather conditions, field of view or quality of cameras.
Far beyond Face Recognition
Face Recognition technology is not only privacy in-compliant, but it works extremely poorly as a standalone solution for CCTV use cases.
Traces AI significantly surpass the performance of Face Recognition for suspect search purpose, guarantying top-notch level of privacy and security to customers' data. While Face Recognition misses tons of information about a person and processing only blurry or occluded faces, our solution detects a vast number of human characteristics from every frame where a person is detected. This is a job that simple facial recognition can't solve. Traces AI captures and analyzes holistic visual information that other systems are unable to detect and process.
Proprietary Deep Neural Networks
Our deep neural networks are developed from scratch, by PhDs in the field of deep learning. We lead the market, providing a new state-of-the-art solution in video security. Traces AI presented the unique approach to solving the problem at NeurIPS (NIPS) Conference - the largest scientific conference in AI. Our team built an algorithm with CCTV use case in mind. It allowed them to develop models that are robust to any angles of cameras, occlusions even bad lightening conditions.
Deep and Lightweight Nets
Our algorithms are lightweight and fast. Traces AI conclusively surpassed all known competitors, thanks to a perfect combination of highly accurate models that operate in real time to keep you and your customers safe. We use fully anonymized data and unbreakable cryptography to guarantee privacy to your data.