Vision Inspection
Computer-vision inspection trained on your brand manual.
Every store walk photo is scored against your standards, with the defect boxed and the manual section cited before the FBC reaches the car.
Inspect a kitchen line photo
Zone score
65AUTO-FAIL
- K1−15
Hot holding below 135°F
FDA Food Code 3-501.16 · Brand Manual §9.3
conf 0.96 - K2−12
Handwashing sink blocked
FDA Food Code 5-205.11 · Brand Manual §9.1
conf 0.92 - K3−5
Missing hair restraint
Brand Manual §10.2 Uniform & Grooming
conf 0.87 - K4−3
Grease buildup under fryer bank
Brand Manual §9.8 Cleaning Schedule
conf 0.81 Sanitizer bucket at 200 ppm quat, test strip logged
Date labels present on all prepped produce
Fryer oil changed per TPM log
What it does in the field.
Exterior & signage
Pylon signs, channel letters, window graphics, parking lot cleanliness and drive-thru menu board illumination.
- Peeling or missing vinyl letters
- Unlit LED modules and channel letters
- Weeds, litter and dumpster enclosures
Marketing & POP
Confirms the current national campaign kit is installed and flags outdated promotional collateral the day a campaign changes.
- Campaign calendar matching
- Window and counter planogram placement
- Unapproved local signage
Merchandising & planograms
Shelf stocking levels, product facing and non-system products on display.
- Facing and gaps
- Unapproved supplier products
- Price label presence
Cleanliness & sanitation
Floor grease, dish stations, overflowing trash and restroom hygiene, with critical food safety items routed to a 24-hour CAP.
- Hand sink access
- Grease and debris build-up
- Restroom supplies and cleanliness
Uniform & grooming
Branded hats, aprons, name tags and food handler hair restraints. Vision AI does no facial recognition; people are detected only as uniform compliance.
- Hair restraints
- Branded apparel
- Name tags
Specifications
- Input
- HEIC, JPEG and PNG from the mobile app, up to 48 MP
- Output
- Bounding boxes, label, confidence, manual citation, severity
- Review
- Findings below your confidence threshold go to FBC review, never auto-scored
- Training
- Tuned per franchise system on your manual and labeled photos; never shared across systems
- Latency
- Median 2.8 s per photo after sync; scored on device for offline previews
Vision Inspection questions
Out of the box it recognizes common franchise defects such as unlit signage, overflowing trash and expired POP. Accuracy on your specific standards improves during the first 30 days as it is tuned on your manual and FBC-confirmed findings. Every finding shows its confidence, and low-confidence detections route to the FBC instead of scoring automatically.
Always. FBCs can dismiss a detection or add a finding the model missed, with a reason. Overrides are logged in the audit trail and used to calibrate your system's model.
No. Gazi Wellness does not run facial recognition. People in a photo are only evaluated for uniform and grooming standards, and you can configure the app to blur faces in exported reports.
No. Your photos, labels and tuned model stay inside your tenant. Gazi Wellness does not send audit photos to third-party AI services.
Double your FBC field audit capacity while eliminating subjective disputes.
Upload 15 store walk photos. See Gazi Wellness score them against brand standards in 60 seconds.
14-Day Free Production Trial on up to 10 stores. Unlimited FBC seats.