478-point face mesh
Full facial landmark tracking runs on-device at camera frame rate, with smoothed boxes, landmarks and expression labels drawn live.
By ProAccess.me
MobiCam
Point your phone at a room and it becomes a CCTV station: every face is found, boxed and tracked with a stable ID at full frame rate, motion can trigger photos or clips on its own, and an optional AI panel reads expression, lighting and photo quality. Export the results as a PDF or a filtered video clip.
Works in the browser no app install, and live video never leaves your device.
Full facial landmark tracking runs on-device at camera frame rate, with smoothed boxes, landmarks and expression labels drawn live.
Each face keeps its own ID while it moves around the frame, so the live counter never flickers and multiple people stay separated.
A slide-up panel sends a low-resolution still to the cloud model and reads back face count, position, mood, eyes-open, lighting and focus.
Frame-difference motion sensing with adjustable sensitivity. When something moves it can snap photos on a timeline or start a recording by itself.
Record with the detection overlay burned in, with 1 min / 5 min / auto-stop limits, chosen FPS, bitrate, resolution and optional audio.
AI auto-capture fires every 5, 10 or 15 seconds and skips blurry or dark frames using live sharpness and brightness checks.
Send captures to the review screen and build a PDF with page size, orientation, quality, margins, footer, page numbers and a watermark.
Re-encode clips to MP4 or WebM with a quality and resolution preset plus colour, grayscale or high-contrast enhancement — then save, share or print.
Open the camera, tap AI Vision and the panel runs these 4 detectors on each polled frame.
Counts every visible face, draws smoothed boxes and landmarks, and reports where each person sits in the frame.
Describes the apparent expression and mood of each face and whether the eyes are open — never names or identifies anyone.
Tells you whether the frame is sharp, evenly lit and clean enough to use as an ID or portrait photo.
Pairs on-device motion sensing with the AI read-out so unattended captures still come with a written description.
1. Tap Start CCTV camera and allow camera access — everything loads straight in the browser. 2. Faces are tracked on-device; switch between the front and back camera, toggle the torch or import an existing photo at any time. 3. Turn on AI Auto Capture or Motion detection to let the camera work unattended, or press the shutter yourself. 4. Tap Finish to open the review screen, then build a PDF from your photos or export your recordings.
The viewfinder is designed for one-handed phone use: large labelled buttons, a shutter that turns green the moment a face is detected, live status pills for tracking and motion, and a pages counter so you always know how much you have captured. Every button, label and size can be reconfigured from the admin console.
Open the camera and the face mesh, motion sensing and AI panel are one tap away — nothing to install.
Open the cameraPrivacy: Video never leaves your device for real-time tracking. Only low-resolution still frames are sent for optional AI analysis. Faces are never identified or named.