Home › Hard problems › India

India: hard problems at India's scale

Reading Indian documents: land records, forms, handwriting

Solo: strongOne person can run a real test of this in about 8 weeks.

खसरा / Khasra No.मालिकOwnerरकबा / Areaसर्वे नं.SEALtwo scriptson one pagestamp over thehandwritingText recognitionbuilt mostlyfor EnglishOwner: ???Area: ???Survey no.: ???
Land records and forms mix scripts, stamps and handwriting, and text recognition for Indian scripts is far behind English, so the key fields come out unreadable.

Why it is still open

Scanned land records and government forms mix scripts, stamps and handwriting. Text recognition for Indian scripts is far behind English.

An experiment one person can run in 8 weeks

Measure text recognition accuracy on 300 public scanned records in 3 scripts across 4 tools, then extract key fields (owner, area, survey number).

Verdict: Solo: strong

Public scans exist, and the field-level accuracy number matters to banks and courts.

The solo test behind the verdict

  1. Can it be measured? There is a number that shows you did better, such as an error rate, a catch rate or a cost per answer.
  2. Is it cheap to try? It runs on a laptop, free open models and open data, or under about ₹50,000 of rented GPU time.
  3. Is there a public yardstick? A benchmark or dataset others use, so your result can be compared and believed.
  4. Would a small win get noticed? A lab, a funder or a customer would care about a 10% improvement, not only a full solution.

Four yeses means Solo: strong. Two or three means Solo: possible. Fewer means Follow, don't build.

PreviousSpeech recognition for dialects and code-mixed talkNextUseful AI on a ₹10,000 phone, offline

All 79 hard problems · Working on this, or know of a result we missed? basant@cleandataengineering.ai