TEACHING MACHINES
TO UNDERSTAND HUMANS

More than 90% of our communication is non-verbal.
This is what makes us unique in the way of expressing and communicating emotions.
We develop artificial intelligence software to sense, assess, and act upon human behaviour.

Request Whitepaper

ROBUST
HUMAN BEHAVIOUR AI SDKS

Our proprietary human behaviour datasets are some of the most advanced in the industry. Using NVISO's massive real-world datasets and AI-powered software it has the capacity to reconstruct a 3D facial model from a 2D image or video, and automatically detects micro-expressions from our existing database of millions behavioral data points to recognize seven universal human emotions, whether it be for one individual in a controlled environment or for a crowd at a live event.

How does it work? NVISO’s visual intelligence technology uses computers to learn from examples opposed to being manually programmed. Using deep Convolutional Neural Networks (CNNs) and state-of-the-art machine learning to understand human behaviors depicted in images and videos, it can achieve accuracy levels that surpass human performance in many narrowly defined tasks.

SCALABLE

CNNs and modern machine learning scale to learn from billions of examples resulting in an extraordinary capacity to learn highly complex behaviors and thousands of categories. Thanks to high volumes of data and powerful computing resources, NVISO intelligence technology can train powerful and highly accurate models.

LOW LATENCY

Our trained models store their knowledge in a single network, making them easy to deploy in any environment. There is no need to store any additional data when new data is analysed. This means that the NVISO visual intelligence engine can run on inexpensive devices with no internet connectivity providing responses in milliseconds.

ACCURATE AND ADAPTIVE

The NVISO visual technology approach does not require human engineered ad-hoc algorithms to extract the discriminative features in an image to make accurate predictions. Our algorithms are able to learn how to extract these meaningful features from the input using only the provided training data. This makes them easy to adapt to problems in any domain and evolve to new capabilities.

EXTREME EDGE ENGINEERING
ULTRA-LOW POWER DEEP LEARNING

NVISO's technology is purpose built for a new class of ultra-efficient machine learning processors for smart edge devices and edge compute with heterogeneous and secure architecture. Supporting a wide range of heterogenous computing platforms ranging from CPU, GPU, NPU, and Neuromorphic computing it reduces the high barriers-to-entry into the AI space through cost-effective standardized AI Apps that are future proof and work optimally at the extreme edge (low power, on-device, without requiring an internet connection).

Support a wide range of activations and weights data types (32-bit floating point to 2-bit binary).
Mixed precision and unstructured sparsity to reduce memory bandwidth and power consumption.
Support for both advanced NN architectures such as RNN, transformers (self-attention), 3D convolution as well as fully sequential architectures for ultra-low power mixed signal inference engines.