Student Team
Kannav Modi
Counterfeit Medicine Team
Panav Golyan
Counterfeit Medicine Team
Mananya Sharma
Counterfeit Medicine Team
A smartphone-based screening system that examines medicine packaging photos and tilt videos to generate a suspicion score, then guides users through safe verification steps.
Counterfeit and substandard medicines are a serious public health problem, especially where consumers and counter clerks have no practical verification method.
Fake medicines are designed to look genuine, with subtle differences across print quality, seals, holograms, labels, and tablet appearance.
The tool never claims a medicine is genuine. It flags suspicious packs for pharmacist, doctor, or drug-inspector verification.
The system uses a medicine-pack photograph and short tilt video to output a calibrated suspicion score, then provides a guided verification and safety information layer.
Compares the query image with a known-genuine reference for the same drug.
Walks the user through visual checks such as label edges, hologram behaviour, seal condition, and date integrity.
Surfaces active ingredient, generic alternatives, expiry status, recalls, and doctor/pharmacist discussion prompts.
Font choice, kerning, character sharpness, ink density, colour fidelity, and print registration.
Signs of label removal, faded edges, residue, and box-edge misalignment.
Blister foil condition, perforations, sealing tape, breaks, or re-sealing evidence.
Short tilt video checks whether holograms show expected colour shifts across viewing angles.
OCR checks over-printing, font mismatch, and visible signs of date alteration.
Colour deviation, mottling, bubbling, visible powder, and consistency across tablets.
Embossed logo or identifier depth, sharpness, and geometry compared to reference.
Consistency of cavity size and shape across the pack and expected product dimensions.
No single visual cue is decisive. ML can combine many small differences better than a human observer.
A Siamese or metric-learning model can compare a query pack against a genuine reference and support new drugs without full retraining.
The deployment device is already common: a standard smartphone camera, without sensors or dongles.
Photographed across brands, manufacturers, batches, and lighting conditions from licensed pharmacies and family inventory.
Uses look-alike products and augmented images with simulated print degradation, colour shifts, hologram absence, and label artefacts.
Where possible, real counterfeit samples can be requested through the District Drug Inspector or State Drugs Control Department.