Logo caractérisation activité
Understand what is happening on an image or video
Analyzing traffic or the number of people in public places, detecting a fall or a person in distress, detecting pedestrians and anticipating when they will cross the street, knowing the occupancy rate of a building or even spotting and moderating inappropriate and violent video content… Humans are able to characterize all these activities, but not without difficulty and errors.
With our activity characterization technology, Neovision assists humans in the analysis of images and videos. This allows them to focus on making informed, fast and reliable decisions. No need to watch hours of video or constantly monitor a multitude of screens, our technology does it for you and helps you optimize your processes.


comptage de personnes

People detection and counting


Traceability is a real industrial issue, becoming more and more important. The Centre Technique du Cuir (CTC) supports French tanneries in their innovation processes. Every day, hundreds of animal skins enter the tanneries and the latter must identify them one by one via unique laser-engraved numbers. Although the characters are often distorted, the OCR technology developed by Neovision makes it possible to read them automatically, quickly and with greater reliability.


Recognition of human activity

Schneider Electric

Building represents 44% of the energy consumption in France(!). It is easy to understand that the energy efficiency of buildings is a crucial issue. With this in mind, Schneider called upon Neovision’s activity recognition technology. The goal is to deduce a type of activity based on the posture of the people present in the building. Depending on the type of activity, the automated BMS adapts the use of the various equipment to reduce or smooth out consumption (turning off the lights, gradually increasing the temperature, etc.).

Caractérisation d'activité
Détection et suivi de piétons

Detection of pedestrians, cyclists and vehicles


Autonomous vehicles elicit fears and fantasies of all kinds. Neovision contributes to the R&D in this field thanks to its activity characterization technology. By merging RGB and infrared images, it is possible to detect humans, regardless of weather and light conditions. Neovision’s technology is able to detect and track the movement of pedestrians, cyclists and vehicles. This is as accurate as it is fast!


Fall detection


Every year in France, approximately 10,000 people die as a result of a fall that is treated too late. The challenge is to detect a fall and alert as soon as possible. Neovision’s activity characterization technology is able to recognize a person who falls, either via infrared or RGB images. Once the fall is detected, an alert is automatically sent to the family who can confirm the problem via the camera.

Détection de chutes
Suivi d'activité et d'affluence

Site monitoring: activity and attendance


How to know the average attendance or the time slots where the activity is the most intense? Most metrics come from Google or from statistics related to recorded sales. But what about people who don’t own smartphones and don’t buy anything? Similarly, how do you monitor the activity of a place like a construction site? Enlaps’ Tikee solution integrates Neovision’s activity characterization technology to detect and qualify the activity captured on timelapses.


Crowd analysis and abnormal behavior


Crowd movements are difficult to understand, track and even more difficult to manage. And panic movements can have disastrous consequences. By analyzing the activity of a crowd, it is possible to anticipate and predict a macro movement. But it is also possible to detect and recognize dangerous behaviors such as confrontations or even weapons and intervene in a targeted and timely manner. The issue is therefore contained avoiding collateral damage.

Analyse mouvement foule
Reconnaissance de contenu sensible et illicite

Detection of violent or illegal content


On Facebook as on Youtube for example, the moderation of video content is a complex subject. While users must be protected from violent and inappropriate content, these platforms expose their moderators to content that is difficult to bear. To assist them but also and above all to protect them, activity characterization can recognize certain behaviors and actions in a video to automatically censor it. AI thus helps protect moderators and users‘ mental health.


    Neovision © 2022