Anvi PanchalAI/ML and Full Stack Engineer
All projects

Project 01, Hardware and computer vision

Coast Scan

A handheld device that analyses beach sand in seconds, instead of days in a lab.

National Winner, Smart India Hackathon 2025, Hardware Edition

How the device sees sand. Each grain is outlined and measured.Illustration
Timeline
Sep to Dec 2025
Team
6 people
Result
National Winner, Smart India Hackathon 2025, Hardware Edition
Now
Incubated at GLS LEAF

01, The problem

Sand tells you a lot about a coast. Measuring it takes days.

Grain size is one of the first things coastal researchers measure. It shows how a beach is changing, where erosion is happening and what kind of sediment is moving.

Today this is done with sieve analysis. A sample is dried, shaken through a stack of sieves and weighed by hand, which takes days of lab work for every sample.

Our problem statement came from the Ministry of Earth Sciences and the National Centre for Coastal Research. They wanted a faster way to get the same answer in the field.

02, What we built

A small lab you can carry to the beach.

Coast Scan is a low-cost handheld device. You put a sand sample inside, and it photographs the grains, measures them and sorts the sediment, then shows the result on a dashboard along with environmental readings.

Black chamber

A closed chamber with its own light, so every photo is taken in the same conditions, on the beach or in a lab.

Camera and sensors

A Raspberry Pi camera captures the grains, and onboard sensors record the conditions around the sample.

Live dashboard

Results appear on a web dashboard with the grain size distribution, the sediment type and the sensor readings.

03, How it works

From a handful of sand to a report.

  1. [ Step 1 ]

    Capture

    The camera takes a close-up photo of the sample inside the chamber.

  2. [ Step 2 ]

    Isolate the grains

    Thresholding, contour detection and morphological operations separate each grain from the rest.

  3. [ Step 3 ]

    Measure

    The device works out the grain size distribution, the D50 grain size and the sediment type. We also tested CNN models for grain recognition.

  4. [ Step 4 ]

    Report

    A Django dashboard on AWS turns the numbers into a report you can read and share.

04, Results

We checked it against the real lab method.

98%

agreement with lab sieve analysis on the D50 grain size

Days to seconds

A report that used to need days of manual lab work is now generated automatically.

Validated in the Civil Engineering laboratory at Sri Sairam Engineering College. The lab report is signed and stamped by the head of the department.

Open the lab report

05, Where it went

From a hackathon idea to an incubated product.

  • National Winner, Smart India Hackathon 2025, Hardware Edition. Selected from 68,766 teams, and the only team from Gujarat among the winners.
  • Design patent filed. "Portable Sediment Grain Size Analysis Device", filed with the Indian Patent Office and under examination.
  • Incubated at GLS LEAF. The incubation centre at GLS University.
  • Pitched at the IDE Bootcamp, Edition 3, Phase 1. Run by the Ministry of Education's Innovation Cell and AICTE at GITAM Bengaluru.
  • Featured in Ahmedabad Mirror. A story on our national win.

Built with

  • Python
  • OpenCV
  • NumPy
  • CNNs
  • Raspberry Pi
  • IoT sensors
  • Django
  • AWS