Industries Telecommunications
Every install photo checked before the technician leaves.
Fibre, fixed wireless and mobile installs are proven with photos. Sight checks each one against what the job requires: the equipment is there, the lights are the right colour, the serial matches the work order.
Antenna Antenna - Antennas on the mount Pass · 4 of 4
- Photo sharp enough to assess Pass · Yes
- Cable labels readable Needs a person · Too far to read
- Check
- Cable labels readable
- Why
- Labels are too small in this photo
- Ask
- Ask for a closer photo of the labels
A stock photo. The boxes are real labels drawn in Sight; the checks shown are an example.
What Sight checks in telecommunications.
Each check is a rule written with you and trained on your own photos. Every answer comes back as pass, fail or needs a person, with the photo and the reason.
The right equipment, in the photo
Each device, panel, antenna and enclosure the job calls for is found and counted. A missing one fails the check, with the photo it was missing from.
Status lights read
Lights on each device are read and compared with what a working install shows. On real install photos, 92 of 92 were read correctly when the device was at least 150 pixels wide. Smaller than that, Sight asks for a closer photo instead of guessing.
Serial numbers matched to the work order
The serial on the label is read and compared with the one on the job, so a swapped or mistyped device is caught the same day.
Every required photo present
A job that needs six photos and has five is flagged while the technician can still take the sixth.
A hint while the crew is still there Rolling out
When a photo is missing or unclear, the person who took it is told what to retake before they leave, not after the job bounces.
Reused photos caught Rolling out
A photo that has already been used on another job is flagged before it is accepted as proof of this one.
In production today on install photos for a national telecommunications contractor. A model trained on those photos scored 0.914 on photos it had never seen.
- 01
Label
We label your photos, with your people confirming what counts as right.
- 02
Train
We train and tune a model on them, and score it on photos it has never seen.
- 03
Set up checks
We write each check with you: what must be in the photo, and what passes.
- 04
Connect
We connect Sight to the app your crews already use and the systems behind it.
- 05
Stay
We keep running it, and retrain it as your reviewers correct it.
You hire no machine learning team and buy no infrastructure. How we deliver
Bring a week of install photos.
An engineer who has built this for telecommunications installs will tell you what Sight can check in them, and what it cannot.