Asked after the mobile-internet question: could a VPN let the office cameras
be shown in the demo. Three jobs get confused there and only one needs one.
Seeing the estate from anywhere already works and needs nothing - viewer mode
plus LiveHub is exactly that, outbound, no installation and no credential.
Demonstrating recognition is better done on the demo machine's own camera.
`webcam: 0` picks a capture index instead of building an RTSP URL and the
engine has supported it since the first version: CameraStore round-trips it,
source() returns the index, safe_url() reports webcam:0, and
POST /api/cameras {"id":"laptop","webcam":0} has always worked. No screen
offered it - the same gap this repo already records for the customer record
and per-camera tuning. It is now an option in the make picker, and it is the
strongest demo available: real faces, in the room, depending on no network.
A demo pointed at a camera in another building depends on two internet
connections and a tunnel staying up while somebody is talking.
The address and the index are alternatives, not extras: source() takes the
webcam first, so a half-typed host left behind would make the saved camera
describe two things and use one. The scan, the path and the camera password
are hidden for a local camera because none of them mean anything.
A tunnel is still right for one case - running the engine on a remote machine
against the office's own cameras - and still wrong for the product: it is
per-site infrastructure on every shop PC, and it gives head office
network-level access into a customer's LAN, where today we can read a
camera's picture and nothing else.
Also: installing httpx took the suite from 239 passed to 271. The HTTP tests
importorskip it so a bare checkout runs, which means the number at the bottom
of a run is not the number of tests that exist. Added to the dev extra.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KGcjxF1cNLcuwc3DAPcnfj
Reported from the first Windows install: the camera feed lags. It did,
and not because of the network, the proxy or the webview.
The MJPEG stream served _annotated_jpeg - the frame the pipeline had
most recently FINISHED with, encoded after detection, quality scoring,
tracking and identification had all run on it. On a modest shop PC that
is a few frames a second, and every picture was already as old as that
processing. It looked like lag because it was lag. On the fast machine
it was developed on the pipeline kept up with the stream's own 10 fps
cap, which is why nobody here ever saw it.
Two more things compounded it. Every processed frame was JPEG-encoded
whether or not a viewer existed - CPU spent on precisely the machine
short of it. And ffmpeg ran its RTSP demuxer with default buffering,
which holds a comfortable queue of frames before handing over the first:
half a second to two seconds a live view can never recover.
Now the picture and the boxes are decoupled. latest_jpeg_since takes the
capture thread's freshest frame at the camera's own rate and draws the
boxes from the last processed frame over it - encoded on demand, per
request, so a camera nobody watches costs no encode at all. The stream
sends a frame only when the camera has a newer one, capped at 15 fps;
nothing is sent twice. Boxes older than a second are not drawn, so a
stalled pipeline cannot leave one floating over an empty spot.
_publish_annotated becomes _remember_tracks: a handful of tuples under
the lock, no copy, no encode. ffmpeg gets nobuffer / low_delay /
max_delay.
Measured on cam2's sub-stream, same machine, ten seconds each:
before 99 frames sent, 98 distinct 9.8 new pictures/s
after 141 frames sent, 141 distinct 14.0 new pictures/s
against a 15 fps camera, with the pipeline still processing 166 of 181
captured frames alongside - and engine CPU DOWN from 90% with no viewer
to 62% with one attached.
Engine version 1.0.0 -> 1.1.0 so a re-run of setup reinstalls it rather
than pip deciding the requirement is already satisfied.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KGcjxF1cNLcuwc3DAPcnfj
Five components that ship as one product:
- behavision/ the recognition engine. RTSP ingest, YuNet detection, IoU
tracking, ArcFace embeddings, a FAISS/SQLite gallery, and a
FastAPI dashboard. Identity is decided once per TRACK from an
average of at least three embeddings, never per frame.
- agent/ the Go edge agent: supervises the engine, holds a durable
spool, and drains it to MQTT. Nothing is acked before the
broker confirms.
- desktop/ the shop PC application (Wails + React + tray).
- server/ the cloud API, MQTT consumer, reports and assistant.
- web/ platform.loyaly.ai, the head-office app, embedded in the
server binary.
The gallery stores 512-float embeddings and timestamps - no images unless
`app.store_faces` is switched on. Those embeddings are biometric personal
data under GDPR and India's DPDP: template inversion reconstructs a
recognisable face from an ArcFace vector, so data/behavision.db is treated
as a biometric database and DELETE /api/visitors/{id} is a real erasure.
CLAUDE.md carries the reasoning behind every non-obvious decision here,
including the ones that were measured and the ones that were wrong first.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HViLj9gYNRtSr7YVZmW5sn
