Audience data for campaigns and storefronts

AI model trained to recognize facial expressions associated with emotions — happiness, sadness, fear, disgust, surprise, anger, contempt and neutral. It measures in real time how customers, students and audiences react to products, service, classes, storefronts and presentations.

How it works

1. 1The YuNet detector locates the faces in each frame.
2. 2The HSEmotion model (trained on the AffectNet dataset) classifies 8 emotions with probabilities.
3. 3Readings are smoothed over time per person to avoid fluctuations.
4. 4A satisfaction (valence) index summarizes the audience's mood minute by minute.

Benefits

• Objective emotional feedback, without questionnaires.
• Identifies the moments and products that generate the most delight (or frustration).
• Complements NPS and surveys with continuous, anonymous data.

Face enrollment and recognition for access control, event check-in, attendance tracking and personalized service. An enrolled face becomes a numerical signature (embedding) compared in milliseconds with every face detected by the cameras.

How it works

1. 1The face is enrolled from a photo or the camera (name and profile).
2. 2The SFace network generates a 128-dimensional vector that represents the face.
3. 3Each detected face is compared with the database by cosine similarity.
4. 4Identified and unregistered people trigger different events (welcome × alert).

Benefits

• Access and check-in without queues, cards or badges.
• Personalized service — the salesperson knows who has arrived.
• Auditable access log.

Anonymous estimation of the audience profile — gender and age range — from cameras. The data feeds audience reports and Personalized Media: screens and kiosks display the content that best matches the profile of whoever is passing by at that moment.

How it works

1. 1Faces are detected and tracked while they remain in the field of view.
2. 2A Vision Transformer (ViT) estimates age in years and gender from frontal faces, several times per person.
3. 3Readings are consolidated per person (median age and mean gender probability), which eliminates fluctuations.
4. 4Each person is counted only once, without storing images or identity.
5. 5The aggregated profile triggers the most relevant media/campaign recommendation.

Benefits

• Know who actually visits the store (not just who buys).
• Campaigns and product mix aligned with the real audience at each time of day.
• Proven audience to sell media space to partners.

Automatic, anonymous people counting: entries and exits through a virtual line, real-time occupancy, unique visitors, average dwell time and a heat map of the most visited areas — turning cameras into an audience sensor for operations and marketing decisions.

How it works

1. 1Each person is detected and tracked with an anonymous identifier.
2. 2Crossing the virtual line defines an entry or exit.
3. 3Accumulated positions form the circulation heat map.
4. 4Reports by hour, day, week and store support staffing, layout and campaigns.

Benefits

• The store's real conversion rate (traffic × receipts).
• Staff scheduling aligned with peak traffic.
• Product layout and display guided by the heat map.
• Capacity control for safety and fire code compliance.