Skip to content

pvbraguinha/smartdog-backend

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

120 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🐾 SmartDog — Biometric Snout Recognition for Dogs & Cats

A computer vision system for biometric identification of dogs and cats using nose print recognition, built to support animal welfare, population control, and zoonosis surveillance under the One Health framework.


🧠 Overview

Animal identification is a critical challenge for municipalities, NGOs, and public health agencies managing stray populations and zoonotic disease risk. SmartDog addresses this by using convolutional neural networks (CNNs) to recognize individual animals through their unique nose prints — a biometric as distinctive as a human fingerprint.

This project was developed as part of Petimuni, a digital platform designed to support municipal animal welfare departments and zoonosis surveillance units in Brazil.


🔬 How It Works

  1. A photo of the animal's snout is captured via the mobile/web frontend
  2. The image is sent to the backend API (hosted on Render)
  3. A CNN model compares the snout against a registered database
  4. The system returns a match confidence score and animal profile

🏗️ Architecture

Frontend (Bolt)
      │
      ▼
REST API (PHP/Laravel — Render)
      │
      ▼
CNN Model (Python — trained on Google Colab)
      │
      ▼
Database + Image Storage (Amazon AWS)

🤖 Machine Learning Model

  • Architecture: Convolutional Neural Network (CNN)
  • Training environment: Google Colab (GPU)
  • Datasets used:
  • Task: Biometric snout matching (identity verification, not breed classification)
  • Key technique: Feature extraction + similarity comparison between snout embeddings

🌍 Real-World Context

This system was developed to solve a concrete public health problem:

  • Brazil has an estimated 30 million stray animals
  • Traditional microchip identification requires physical capture
  • Municipalities lack scalable tools for population tracking
  • Uncontrolled animal populations are a primary vector for zoonotic diseases (rabies, leishmaniasis, leptospirosis)

SmartDog enables non-invasive remote identification using only a photograph, making it viable for field use by animal control agents and veterinary teams.


🚀 Deployment

Component Technology
Frontend Bolt
Backend API PHP · Render (403 deployments)
ML Training Python · Google Colab
Storage & Infrastructure Amazon AWS
Database MySQL (Replicate integration)

📁 Repository Structure

smartdog-backend/
├── app/              # Core application logic
├── bootstrap/        # App initialization
├── config/           # Configuration files
├── database/         # Migrations and seeders
├── public/           # Public assets
├── resources/        # Views and templates
├── routes/           # API route definitions
├── storage/          # File storage
└── tests/            # Automated tests

👨‍💻 Author

Paulo Victor Braga de Almeida Santos
Veterinarian | Startup Founder | Aspiring AI Researcher
LinkedIn · pvbraguinha@gmail.com
Coimbra, Portugal


🔗 Related Work

  • Phlebotomic fauna and seroprevalence for canine visceral leishmaniasis in an urban area in the Midwest region of Brazil — Arquivo Brasileiro de Medicina Veterinária e Zootecnia, 2020
  • Epidemiology of Norovirosis and Study of the Role of the Dog as a Reservoir for this Zoonotic Agent — UNICIÊNCIAS, 2019

"The protection of human health begins with the health of animals and the environments we share."
— One Health Initiative

About

Biometric snout recognition API for dogs and cats using computer vision — One Health framework

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors