Skip to content

Easarc VisionAI quality inspection

Catch the defect on the line, not in the buyer's warehouse.

Mount an ordinary IP camera. Vision flags fabric defects, print misalignment and packing errors in real time, and tells you which shift and which machine they came from.

The inspection view

A defect, the moment it happens, with its address attached

Knowing that 2.4% of the lot is defective is not useful. Knowing that it started at 14:20 on machine PR-04 after the shift changed is what lets you go and fix it.

Easarc Vision — Live inspection
CAM-03 · Printing
CAM-03 · Machine PR-04 · Shift B
Live
Weft bar
Oil stain
08 Aug 2026 · 14:21:06 IST

2 defects in the last 60 seconds · linked to ESR-2026-0431

Defect mix, last 24h

  • Weft bar34
  • Print misalignment28
  • Oil stain21
  • Shade variation18
  • Hole / cut12

By shift

ShiftOutputDefect
A41,200 m1.8%
B38,600 m2.4%
C27,400 m3.6%
Threshold crossed

Shift C is at 3.6% against your 3.0% limit. Supervisor notified on WhatsApp at 06:12.

A
Every frame carries the camera, the machine and the shift. That is what turns a defect count into something a production head can act on.
B
Defects attach to the work order running on that machine at that minute, so the QC report for a buyer builds itself.
C
The taxonomy is yours. A process house and a garment unit do not have the same defects, and a shared list helps neither.
D
Rates broken out by shift are usually the first uncomfortable thing the system shows you, and the first thing worth fixing.

How the pipeline runs

Capture at the edge, infer in Mumbai

Cameras push frames to a small edge box on your LAN. It buffers, samples and forwards — so a dropped internet link delays analysis instead of losing footage.

  1. Capture

    Any ONVIF or RTSP camera. Frames are sampled at the rate your line speed needs, not at whatever the camera happens to emit.

  2. Buffer at the edge

    AWS IoT Greengrass on a small box in your unit holds frames when the link drops, then drains the backlog when it returns.

  3. Infer in the cloud

    A SageMaker endpoint in ap-south-1 runs your tuned model. Round trip is typically under two seconds from frame to flag.

  4. Act

    The defect lands on the supervisor’s phone, on the work order in Flow, and in the daily QC report your buyer will ask for.

What you get

Everything Vision does

  • Custom defect taxonomy per customer
  • Edge capture with cloud inference
  • Shift, operator and machine analytics
  • Alerts when defect rate crosses your threshold
  • Exportable QC reports for buyer audits

Integrations

It has to work with what you already run

Nobody replaces their whole stack for one vendor. These are live today, and the API covers the rest.

ONVIF / RTSP cameras
Any IP camera that speaks RTSP — no proprietary hardware
AWS IoT Greengrass
Edge capture and buffering when the line drops
Amazon SageMaker
Per-customer model training and tuning
Easarc Flow
Defects attach to the work order they came from
Slack / WhatsApp
Threshold alerts to the supervisor on duty
S3 export
Clip and report archive in your own bucket

Pricing

Easarc Vision plans

Every paid plan starts with a 14-day free trial and no card. Annual billing gives you two months free.

Starter

₹4,999/camera/mo

Billed monthly

  • 1 camera
  • Standard defect models
  • 30-day clip retention
Start free trial

Growth

Most chosen

₹12,999/mo

Billed monthly

  • Up to 5 cameras
  • Custom defect taxonomy
  • Shift analytics
  • 90-day retention
Start free trial

Enterprise

Contact sales

  • Unlimited cameras
  • On-site model tuning
  • Buyer-audit reporting
Contact sales

GST extra at 18%. All paid plans include a 14-day free trial, no card required.

USD shown for reference at ₹88 = $1. Invoices are issued in Indian rupees.

Compare every plan feature by feature

Get started

Put Vision in front of one line

You do not have to commit the whole unit. Run Easarc Vision against a single product line for a fortnight and judge it on what changes.

14-day free trial, no card required. GST extra at 18%.