App 02 / Missing children

Be the eyes when her face matches your route.

A child is reported missing every four minutes in India. Most are girls. Police FIR data is isolated; one stranger's snapshot at a traffic light, validated by a trained reviewer, becomes a national distributed sighting network.

A parent's hands holding a printed photograph; the child's face is not shown, protecting identity.

The size of the gap

The state already has the photos. The state cannot have eyes everywhere.

Khoya-Paya, the government's missing-children portal, has existed since 2015. Adoption is limited. The technical question was never 'is this possible' — it was 'why is it not built'.

8.68 lakh

people went missing in India in 2023. 4.07 lakh remain untraced. The state's existing infrastructure cannot reach the moment a face is seen at a traffic light.

Source: NCRB Crime in India 2023

Every 4 min

a child is reported missing in India — roughly 1.4 lakh per year. About 49,000 are still untraced as of NCRB 2023.

Source: NCRB Crime in India 2023

71%

of untraced missing children in India are girls. Most are between the ages of 12 and 18. The category most exposed to trafficking and forced labour.

Source: NCRB Crime in India 2023

The mechanism

Take the photo. Chakshu does the rest.

  1. 01

    A parent (or an NGO) registers a face.

    One clear photo plus consent. The image becomes a vector embedding inside an encrypted store. The original photo never leaves a private bucket.

  2. 02

    Volunteers scan in passing.

    A volunteer sees a face that gives them pause — at a traffic light, a station, a beach. One tap on Chakshu, one frame, no upload. The match runs locally and sends only a hashed signal up.

  3. 03

    A trained NGO worker reviews.

    A reunion never happens on a confidence score alone. Candidate matches go to a human reviewer at our partner NGO who triggers the next step — police FIR, child welfare, or family contact.

The ethical frame

Face-match is the easy part. Consent and review are the hard parts.

  • Consent on the record.

    Every photo we hold has documented parental or guardian consent. No scraping. No public-dataset ingestion. No partner is allowed to bypass this.

  • Human in the loop.

    Algorithms suggest. Trained NGO workers verify. A reunion never happens on a confidence score alone.

  • Right to be forgotten.

    Once a child is recovered, the family can request immediate deletion of the embedding and the original image. The deletion is logged and verifiable.

  • No commercial use.

    The face-match infrastructure is restricted to verified NGO and law-enforcement partners working on missing-children cases. There is no API tier for anyone else.

“A kid at a red light begs at your car window. Your phone has a camera, GPS, and access to a face-match backend. The technical question was never whether this is possible. It was why no one had built it.”

Ashish Sharma, Founder

Volunteer pool

Join the Chakshu pilot.

Open to parents and families currently searching, and to NGO partners working in child rescue and reunion. We onboard slowly so the consent and review processes hold.

One signup. Three apps. We never sell or share. Unsubscribe in one click.

The volunteer pool

One signup. The same network powers SecureMe and Sevarthi.

Chakshu volunteers double as SecureMe responders and Sevarthi givers. The right responder is always the closest one.

Read the volunteer brief