Buddy punching, edited timesheets, and staff clocking in from the car park are attendance frauds that quietly inflate payroll every month. This build-to-order module replaces the honor system with proof: an AI face match plus a GPS coordinate captured on every single punch, so a check-in only counts when the right person is physically inside an authorized location.

ECOSIRE designs, builds, installs, and supports this module for your Odoo 17, 18, or 19 database (Community or Enterprise). It is NOT an instant download from apps.odoo.com — it is a net-new module we engineer to your locations, shift rules, and privacy requirements, then hand over as clean, documented source you own. Typical delivery is one working week from confirmed scope and access.

Under the hood we extend the standard hr.attendance model rather than replacing it, so every punch flows straight into Odoo's native Attendances, Timesheets, and Payroll reporting with no parallel system to reconcile. New fields (face_match_score, gps_latitude, gps_longitude, geofence_id, selfie_image, ip_address) are added via a proper models.Model extension, with a computed is_within_geofence flag driven by @api.depends on the captured coordinates and the assigned geofence radius. A new hr.geofence model stores each authorized location as a center point plus radius (or polygon), and record rules scope employees to only the geofences their department or contract allows.

The face-recognition engine runs against enrolled reference photos and returns a confidence score on each attempt; you set the acceptance threshold, and low-confidence or off-site attempts are flagged (or blocked) and routed to a manager for review via an automated action. Employees check in from a mobile browser/PWA on their phone (GPS + selfie from the device) or from a fixed kiosk tablet at the entrance. Every record stores the selfie, the score, the coordinates, and the request IP address for a complete, tamper-evident audit trail.

Managers get a geo-map dashboard of live punches, exception lists for out-of-fence attempts, and QWeb PDF/Excel attendance reports filtered by geofence, employee, or date. Because the module speaks Odoo's own XML-RPC/JSON-RPC API, kiosks, gate hardware, or an external HR app can post attendance programmatically. We build it around your consent and data-retention rules so biometric handling is deliberate, not accidental.

You receive the full module source, an install on your server (staging then production), configuration of your real geofences and thresholds, admin training, and a support window. This is proper Odoo engineering — security CSV, record rules, tested views, upgrade-safe manifest — not a black box.