Cameras in Rice Mills and Agro-Processing Units (India)

In a rice mill, cameras earn their keep at three points: the weighbridge where paddy arrives, the despatch gate after dark, and the milling section where out-turn quietly slips. On a 4 TPH mill handling 20,000 tonnes of paddy a year, half a percentage point of lost out-turn is roughly ₹30 lakh — several times the cost of the cameras.

If you own a mid-size rice mill, dal mill, flour mill or agro-processing unit — 200 to 1,000 workers across milling, drying yard, godown and despatch, in Karnal, Nizamabad, Raichur, Burdwan or Chhattisgarh's Mahasamund belt — you run one of the most common plant types in Indian manufacturing and one of the least written about. Tens of thousands of rice mills operate across the paddy states, yet almost every CCTV article assumes an auto ancillary in Pune.

Key points

• Your out-turn obligation is a fixed number, and it is unforgiving. Under the custom-milling policy, a miller is required to deliver 67% rice, including 1% Fortified Rice Kernels, against the allotted paddy, on a staged delivery schedule. Every kilo short is bought back at market cost.
• Replacement rice is dearer than the paddy you lost it from. Paddy MSP for KMS 2025-26 was fixed at about ₹2,369 a quintal for common and ₹2,389 for Grade A, while open-market rice from central stocks has recently been priced around ₹3,000 a quintal. A 100-tonne shortfall is roughly ₹30 lakh to make good. Check the current MSP and OMSS reserve price before you use these numbers in your own model.
• Moisture is where the gate argument starts. Central procurement specifications cap paddy moisture at 17%. Wet paddy accepted at the gate becomes your drying cost and your out-turn loss — not the farmer's or the arhtiya's.
• Night despatch is the leak owners discover last. One unrecorded 15-tonne truck a quarter is 60 tonnes a year — roughly ₹18 lakh at replacement value. See gate and weighbridge pilferage for the mechanics.
• Dust will kill badly chosen cameras, and IP ratings are less protective than the box implies. Under IEC 60529, IP5X means "dust protected" and IP6X means "dust-tight" — but a sealed housing does nothing for a caked lens or a milky dome bubble. Budget cleaning labour, not just hardware.
• The Factories Act framing is dead. India's four labour codes came into force on 21 November 2025 and the OSH (Central) Rules, 2026 were notified on 8 May 2026. Your safety file must reference the OSH Code, 2020, not the 1948 Act.
• A camera system for a mill of this size is small money. At ₹8,000–₹15,000 per camera installed plus ₹1,000–₹2,500/camera/month for AI, a 24-camera mill lands near ₹2.5–3.5 lakh upfront and ₹5–7 lakh a year.

Why a rice mill is a different camera problem

Three things make a mill unlike a machine shop.

It is seasonal and it runs flat out. During the milling season the plant may run 20 hours a day for weeks, with contract labour, hired trucks and unfamiliar drivers moving through the yard. Supervision thins exactly when volume peaks.

Its product is fungible. A gearbox casting is traceable. A quintal of rice is not. Paddy, bran, broken and husk all have an open market, and none of them carries a serial number.

Its accounting is a physical mass balance. Paddy in, rice out, plus bran, broken and husk. If the balance is short, the miller pays. That means the two things worth watching are what crossed the boundary and what happened between the two weighments.

The four places money actually leaves a mill

1. The gate and the weighbridge. Short weighment inbound, inflated tare, an axle off the platform, a truck weighed twice, a lorry that entered but was never billed. This is the highest-value camera position in the plant.
2. The drying yard and paddy stack. Over-dried paddy loses weight and breaks in milling; under-dried paddy spoils and drags out-turn down. It is also the easiest place for a stack to be quietly short.
3. The milling section. Huller and whitener settings, rubber roll wear and blend changes move broken percentage by shift. Nobody logs when the setting changed or who changed it.
4. Despatch after dark. The single loading bay, at 11 pm, with no supervisor and no gate register entry.

Camera map for a 4 TPH mill

Twenty to twenty-eight cameras cover a 4 TPH mill properly. The placement logic is set out in more detail in where to place cameras on a shop floor.

Out-turn, shift by shift: what a camera can and cannot do

Be precise here, because this is where vendors oversell.

A camera cannot measure your out-turn. Out-turn is a weight ratio. It comes from a weighbridge and a bagging count, not from a lens.

What a camera can do is give you the shift-level context that explains the ratio. When you put the day's out-turn next to a shift log of machine running time, stoppages, blockage clearances and who was at the panel, the pattern usually appears within a couple of weeks: one shift consistently a few tenths lower, one operator's setting change, one dryer batch pushed too hard. Owners typically get this as a single end-of-shift summary rather than by watching video — the format described in the daily shift report.

Independent bag counting at bagging is the other honest win: a machine count off the stitching line, compared against the despatch note, is a real control that does not depend on anyone's memory.

The rupee model: 4 TPH mill

Assume 4 tonnes of paddy an hour, roughly 20,000 tonnes a year, at an MSP-linked value of about ₹23,690 a tonne — around ₹47 crore of grain crossing the gate. Replacement rice is taken at ₹30,000 a tonne. Substitute your own current prices; the shape of the answer does not change.

You do not have to recover all of it. Recovering a fifth of it pays for the system more than once. The general payback logic sits in does camera AI pay for itself.

Dust: the honest section

Rice milling produces husk and bran dust continuously. Two vendor claims deserve scepticism.

"IP66 means dust-proof, so it will be fine." IP6X means the housing is dust-tight — no ingress into the electronics. It says nothing about dust on the front glass, and a dome bubble in a milling hall can go from clear to milky in a week. Prefer bullet or turret bodies with recessed glass over domes inside the milling section; domes are fine in the office corridor.

"Install it and forget it." Plan a wipe-down schedule — weekly in the milling and bagging sections during season, monthly elsewhere — and put it in the AMC in writing. Keep cameras out of the direct discharge path of aspiration and cyclone outlets, and away from dryer exhaust, where heat plus fine dust bakes onto glass.

Dust is also a safety issue, not just an optical one: grain dust and hot work are a well-documented fire and explosion combination in mills. Cameras will not prevent that — hot-work permits, aspiration and housekeeping will. What a camera adds is a record of whether hot work happened where it should not have.

Compliance: what cameras genuinely help with

• Procurement and custom-milling audits. When a state agency or FCI questions receipts, deliveries or a shortfall, timestamped weighbridge and despatch footage plus the gate register is a far stronger file than a ledger alone. Set retention to cover the whole delivery window — not the 15 or 30 days most NVRs ship with.
• FSSAI. Milling is food manufacturing, so you carry a state or central licence depending on turnover, and hygiene obligations apply. There is no blanket FSSAI CCTV mandate — see food processing CCTV and FSSAI.
• OSH Code, 2020 and the OSH (Central) Rules, 2026. Camera evidence supports incident investigation and contractor-labour headcount, but it substitutes for nothing statutory. There is still no general legal requirement to install CCTV in an Indian factory. The real exception is night work by women: the central rules and several state conditions require surveillance coverage as part of the safeguards for employing women outside normal hours. If you run women on night shifts, that is a genuine obligation — check the version applicable in your state.
• Buying the hardware. Certification requirements now apply to CCTV equipment sold in India. Ask the vendor for test and certification documentation for the exact models supplied, and do not buy grey-market cameras for a plant that gets audited.

Where cameras do not help

Say this plainly to any vendor who claims otherwise:

• They cannot verify moisture or grain quality. A 17% moisture call comes from a meter and a sample, not a lens.
• They cannot detect software-side weighbridge tampering. If the indicator is manipulated electronically, the picture looks perfectly normal. Only the weighbridge's own audit trail and legal-metrology verification catch that.
• They cannot count grain. Bags, yes. Loose paddy in a stack or heap, no — volume estimation from video in a dusty yard is not reliable enough to argue with.
• They cannot fix a milling problem. Rubber roll wear, poor paddy separation and blend variation are engineering issues. Cameras only tell you when things changed.
• They will not stop collusion between the gate and the office. They make it slower and riskier, which is worth a lot — but the control that works is reconciliation, not watching. The same point applies to internal theft generally.

FAQ

How many cameras does a 4 TPH rice mill need? Typically 20–28: two on the weighbridge, two at the gate with ANPR, one or two per milling line, one at bagging, two to three at despatch and the godown door, plus drying yard and stack coverage. Fewer cameras in the right positions beat forty pointed at nothing in particular.

Will cameras improve my out-turn percentage? Not directly — out-turn is weight, not video. What they give you is shift-by-shift context: machine idle time, stoppages, panel changes and independent bag counts. Owners who review that alongside daily out-turn usually find one or two recurring causes within a fortnight.

What IP rating should I specify for a dusty mill? IP66 as a floor for the milling hall, drying yard and any outdoor position. But the rating protects electronics, not image quality — the practical requirements are recessed front glass, positioning away from aspiration and dryer exhaust, and a written cleaning schedule in the AMC.

How long should a rice mill keep footage? Long enough to cover the custom-milling delivery window and any procurement dispute — a season rather than a fortnight. That is a storage decision made at design time, and it is far cheaper to size the drives correctly on day one than to retrofit later.

Do we need CCTV to satisfy FCI or state procurement agencies? No general rule compels it for milling itself. But when a shortfall or a receipt is disputed, weighbridge and despatch footage tied to gate-pass numbers is the difference between arguing and showing. Treat it as evidence infrastructure, not compliance — and note separately that women's night-shift rules do carry surveillance conditions.

Can this run on the cameras we already have? Often partly. Existing analogue or basic IP cameras at the gate may be reusable for context, but weighbridge and ANPR positions almost always need proper WDR and IR to be evidentially useful. Audit what you have before buying anything new.