HUPI – Cyber-Physical System for Detecting People and Vehicles Using Computer Vision
HUPI
Sector: Mobility
Business Case
This system offers 3 main functionalities, ALL in real time: (1) Tracking a person, vehicle, animal, etc., and following them in a delimited environment; (2) characterising and counting such persons or vehicles present in that environment; and (3) predicting future influx at specific locations.
Objectives
Determine the influx and behaviour of visitors/vehicles in real time, improve the reception of visitors/vehicles, adjust resources (human and material) to the influx, optimise spaces and overall performance.
Use case
All videos are processed locally in the systems deployed in the Hupi environment to extract the figures (volume of vehicles, type, locations). The videos are NEVER accessible to anyone and are automatically destroyed once processed = 100% confidentiality. The numbers ARE THE ONLY thing that is sent automatically and in real time to Hupi’s central platform for analysis, processing and visualisation.
Infrastructure
Cloud
Technology
Image recognition/processing
Data
Inflow history, map of the delimited physical environment, OPEN data (weather, specific events that affect the level of influx in the environment).
Resources
Human resources: multidisciplinary team: 1 technical director, 2 mathematicians & data scientists, 1 full stack developer, 1 front end developer and digital UX expert team expert in the domain. Hupi Elastic Cloud Platform
Difficulties and learning
InInstallation of systems in natural environments: lack of electricity, need for connection to the grid, etc. Extreme climatology: maximum heat and humidity during the summer months. Communication with the data visualisation platform.
KPIs (business impact and metrics of the model)
Reduced queues and overcrowding. Improved personnel management. Waiting time in specific environments.
Funding
Private.
Collaborators, Partners
Own resources
