AI Based Pet Adoption

AI-based Pet Adoption System

Sumit Kajbaje

Rohit Sawant

Ronit Loke

Vishwajeet Patil

Our Goal

Our goal is to develop a universal animal recognition to increase the knowledge about the global biodiversity

Motivation

To adopt a pet is not an easy decision.
Many people are afraid to adopt a pet because they do not want to take care of them.

Problem Statement

Millions of dogs and cats are currently in shelter and rescue care center, waiting to be adopted.
Create a mobile application that replaces the current time-consuming pet adoption process with a mobile friendly one.
An adopter can search pets according to distance, gender, breed, age an weight, along with behavioral characteristics.
Adopter can also easily contact the shelter/rescue care.
Adopting them is a very long and rather a slow process.

As per WHO, there are 200 million stay dogs and 3.3 million of them enters the animal shelter every year.

Only 2% - 17% of pets returns to their owners. Due to lack of space in shelters the rest of the animals are put to rest (killed)

There are over 100 diseases a human can catch from an animal. 40,000–70,000 People Die From Rabies Each Year.

The current methods used to adopt or looking for pet is inefficient and haphazard. People distributed flyers to public and spread the news to the people around the area where the pet is lost. When people print out flyers, it results into waste of paper and money as it does not have any positive outcome. People also post stories on social media platforms such as instagram, facebook. There are numerous scammers too who try to falsely claim the reward for returning the pet to their rightful owner.

People are turning to adoption apps because they don't have the time to make frequent stops at animal shelters. To find their next companion they need a reliable and intuitive app, but find themselves frustrated with the lack of features and information, leading to no adoption.

Literature Survey

Title	Year	Author	Description
Bird Species Identification Using Image Mining & CNN Algorithm	2020	Saundarya Junjur Punam Avhad Deepika Tendulkar	Had higher accuracy but with restricted access.
JSP-Based Pet Adoption System	2019	Haoran Liu Xiue Meng	Basic logic of Inventory Management System, lack of information, Not secured

Title	Year	Author	Description
Individual Cattle Identification Using a Deep Learning-Based Framework	2019	Yongliang Qiao Daobilige Su He Kong Salah Sukkarieh Sabrina Lomax Cameron Clark	Has more precision but only limited to cattle for identification. Practical only on a small to medium scale.
Scrutiny of Methods for Image Detection and Recognition of Different Species of Animals	2019	Elham Mohammed Thabit A. ALSADI Nidhal K. El Abbadi	Highlights the different algorithms that can be used for image recognition.

Title	Year	Author	Description
Animal Identification	2018	Jayati Bodkhe Harshita Dighe Aparna Gupta Litesh Bopche	Needs to assign RFIDs on individual animals. Provides great security and doesn't get affected by the environment. Costly to deploy and might harm animals

Existing System

Contacting animal shelter and visit them.
Filling out an application.
Choosing your new pet.
Experiencing the waiting period (24 - 48 hrs)
Signing a contract and paying a fee.
Undergoing a trail period.

Proposed System

A System to...

Find potential adopters.
Make adoption process smoother.
Providing information to make informed decision.
Option to donate money to rescue shelters to support their cause
Easily schedule an appointment.
Create seamless experience for adopters so that they can only focus on finding their right pet.

System Architecture

The image recognition algorithm (image classifier) takes the image (or a patch of the image) as input and outputs what the image contains. In other words, the output is a class label (fox, wolf, bear etc.).

The animal recognition and classification system

User-side Flow Diagram

Admin-side Flow Diagram

Animal Recognition System

Pre-processing Block
Feature Extension Block
Classification

Modules

User Authentication, Authorization & Role Management.
Recognizing an Animal based on an Image.
Showing Adoption Centers & Vets nearby
Verifying the user's Emotional and Mental health.
Pet Recommendation based on user's lifestyle.

Algorithms

Convolutional Neural Network
K-Nearest Neighbor
Pattern Matching
Sentiment Analysis
Content-based Filtering

Convolutional Neural Network

Classification Process

Foreground-background segmentation and object verification

Detects animals in highly cluttered videos

Ensemble graph cuts for object classifier as foreground-background segmentation followed by object verification

Precision: 0.82
Fusion of deep learning and HOG features

Detects animals in highly cluttered videos

Precision: 0.82
Deep CNN with machine learning

Detects animals in highly cluttered videos

Precision: 0.91

Architecture of CNN for Animal Detection

Block Diagram of Animal Detection System

CNN is mainly constructed by using following layers:

Convolutional Layer
Pooling Layer
Flattening Layer
Fully Connected Layer

Architecture of CNN for Animal Detection

K-Nearest Neighbor

Finding Nearest Points

Comparison Algorithm

The k-nearest neighbors (KNN) algorithm is a simple, supervised machine learning algorithm that can be used to solve both classification and regression problems. It's easy to implement and understand, but has a major drawback of becoming significantly slows as the size of that data in use grows.

KNN Algorithm

Select the number K of the neighbors
Calculate the Euclidean distance of K number of neighbors
Take the K nearest neighbors as per the calculated Euclidean distance.
Among these k neighbors, count the number of the data points in each category.
Assign the new data points to that category for which the number of the neighbor is maximum.
Our model is ready.

Simple & supervised Algorithm.
Used to solve classification and regression problems.
Easy to implement & understand.
Drawback: Slows down as the size of the data increases.

Pattern Matching

To find a pet that suits your lifestyle

Sentiment Analysis

To know the opinion.

Sentiment analysis is a predominantly classification algorithm aimed at finding an opinionated point of view and its disposition and highlighting the information of particular interest in the process.

Algorithm Scope

Document-level: entire text
Sentence-level: for a single sentence.
Sub-sentence level: for sub-expressions, words.

Types of Opinion

Direct: Directly stated.
Comparative: opinion that compares two things.
Explicit: clearly defined.
Implicit: implemented but not clear.

Why Matters

Customer support correspondence (regarding your product)
User-generated Product reviews
Professional product reviews (as in The Verge or Wired)
Social Media tractions
General and special-purpose forums

Content-based Filtering

Recommendation System

Content-based Filtering
1. Find out how many users/items in the database are similar to the given user/item.
2. Assess other users/items to predict what grade you would give the user of this product, given the total weight of the users/items that are more similar to this one.
Content-based Filtering

Alternative Algorithms

K-Nearest

Simple & supervised Algorithm.
Used to solve classification and regression problems.
Easy to implement & understand.
Drawback: Slows down as the size of the data increases.

Comparison Algorithm

KNN Algorithm

Select the number K of the neighbors
Calculate the Euclidean distance of K number of neighbors
Take the K nearest neighbors as per the calculated Euclidean distance.
Among these k neighbors, count the number of the data points in each category.
Assign the new data points to that category for which the number of the neighbor is maximum.
Our model is ready.

Amazon's A9

Developed by Amazon.
Used to decide the ranking of the products.

Search Algorithm

The A9 Algorithm is the system which Amazon uses to decide how products are ranked in search results. It is similar to the algorithm which Google uses for its search results, in that it considers keywords in deciding which results are most relevant to the search and therefore which it will display first.

Amazon A10 algorithm aims to increase visibility for listings with authentic reviews and sales. Apparently, it's trying to find out fake profiles and address the major problem of counterfeits and fake reviews.

Mathematical Module

Advantages

Process that is simpler, easier, saves time, is clearer, transparent, trustworthy, and more effective.
Provides detailed knowledge.
Reduces Cost and Human efforts.

Disadvantages

Need large number of sample data for getting more accurate results
Recognition results can never be 100% accurate.
Data to be store may increase drastically based on the region it covers.

Applications

Application can be used by

Individual
Shelter services
Pet Stores
Veterinary surgeons
Animal Rescue Service & Organizations

Conclusion

The suggested method improves analytical accuracy, minimizes human labour, and makes the adoption process more useful, faster, and effective.

References

Stray Animal StatisticsーWhat Are the Actual Numbers in 2021
A research on Animal adoption by Research Gate
Prevention of cruelty to animals Act 1960 Prevention of Cruelty to Animal, Wikipedia
Bird Identification by Image Recognition. Madhuri A. Tayal, Atharva Mangrulkar, Purvashree Waldey, Chitra Dangra, 2018.
Bird Species Identification using Deep Learning. Prof. Pralhad Gavali, Ms. Prachi Abhijeet Mhetre, Ms. Neha Chandrakhant Patil, April-2019.
Large-Scale Bird Sound Classification using Convolutional Neural Networks Stefan Kahl, Thomas Wilhelm-Stein, Hussein Hussein, Holger Klinck, Danny Kowerko, Marc Ritter, and Maximilian Eibl