Tray-return cameras that show campus kitchens which dishes come back uneaten
TL;DR Leftoverlens puts a scale and an overhead camera at the tray-return counter of corporate cafeterias, college messes and hospital canteens. It shows the caterer, dish by dish, what people leave on their plates. Caterers pay a monthly fee and recover it by cooking less of what nobody finishes.
Indian kitchens cook by habit. In a survey of 63 Mumbai restaurants, three in four deliberately made 10–20% more than needed, and some went as far as 30%[1]. Across foodservice, the leftovers on plates can be the biggest slice of waste[1].
Hardly any kitchen measures it. A review of university food waste found 60% of 52 institutions surveyed didn't track canteen waste at all[2]. When one Indian canteen did act, its waste fell by half, though that study also covered COVID-era changes[2].
Here's the interesting bit - Indian students were 40.5% more likely than UK students to say they left food because it didn't taste good[2]. That's a menu problem, and a caterer can fix it if it can see it.
Leftoverlens sits where diners return trays. A load cell under the scraping bin weighs whatever gets tipped in, and a camera above the drop point photographs each plate before it's scraped. Software matches the leftovers to that day's menu and reports by dish, by meal and by counter, so you see that Tuesday's rajma came back 30% uneaten and the new pasta 45%.
The camera frames only the tray and never faces, and it keeps no images after processing. Image models are far from perfect on mixed plates, and a 2023 review found error rates from near zero to 38% for calorie estimates, with simple foods easiest[3]. So Leftoverlens takes weight from the scale and uses the camera only to split that weight between dishes. Weekly reports tell the chef which recipes to change and how much less to cook.
The goal is the original one, but the first customer is narrower. Restaurants rarely have a single return point, while cafeterias, hostels and hospital kitchens do.
Contract caterers running corporate cafeterias and college messes pay, because food cost is their margin. In some contracts the client company's facilities team pays as part of a sustainability goal. A test price of ₹6,000–12,000 per counter a month including hardware is an assumption you'll need to check.
One UK competitor says its kitchens save US$5,000–50,000 a site each year, or roughly ₹4.8–48 L[4]. That's a vendor claim, but it gives you a ceiling.
FSSAI started enforcing its surplus-food regulations in 2019, which nudges large kitchens to account for what they cook[1]. Cheap cameras and edge processors make a counter unit affordable now. Global vendors have already sold the model to hotel chains and contract caterers abroad[4][5], so Indian buyers can check references before they sign.
Nobody publishes a figure for India's institutional catering spend or for plate-waste tech. The tightest evidence is about behaviour - most institutions don't measure waste[2], and Mumbai restaurants overcook by 10–20% on purpose[1].
Purely as context, Indian households waste about 68.7 million tonnes of food a year, roughly 50 kg each[6].
UK bin-camera firms sell bin tracking and production forecasting to global hotel chains and contract caterers, and claims it cuts waste by about half[4]. A US vendor sells kitchen-waste and plate-waste tracking to caterers, universities and hospitals[5]. Neither mentions India on its homepage.
I couldn't find an India-built plate-waste product, though web search was limited during this research. The gap is a cheaper unit tuned to Indian thalis and mixed plates.
| Difficulty | Moderate (48/100) |
| Starting capital | ₹55 L–1.1 Cr |
| Time to launch | 6–9 months |
The hardware is simple. Telling dal from sambar from kadhi on a steel plate isn't, and you'll need thousands of labelled photos. Selling is slow too, because caterers have to let you into client sites and staff can't skip the counter at rush hour.
No sector licence for tracking plate waste. The rules that apply cover your hardware and the images you capture.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't trust the camera's dish split on mixed plates without the scale, since error rates vary widely[3]. And never point it at faces.
Want to work out the AI layer for Leftoverlens? Talk to Yash →
Here's what can go south -
I like this for a founder with computer-vision skills and a contact in contract catering. If you're hoping the hardware will sell itself then it won't.
Prove this first, in two cafeterias over six weeks - that dish-level reports make chefs change quantities, and that food purchases fall by at least 5%.
Plan for roughly ₹55 L–1.1 Cr, covering scale and camera units, software, labelled photos for training and pilots at a few sites.
No sector licence. You need WPC self-declaration for wireless counter units, DPDP Act compliance for images and the usual company and GST registration.
Yes. Cafeterias everywhere have the same problem, and global vendors already exist. Check your local data protection and wireless equipment rules first.
The free Leftoverlens validation kit has the line-by-line capital breakdown, the effort scoring, the radio and data-protection checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for caterers and facility heads, and 5 assumptions to test. Download the kit →
Thinking about building Leftoverlens or something close to it? The make-or-break work is getting a contract caterer to share the savings and not hide the waste, and that's a conversation I help founders prepare for. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.