Recognition of Bird Species Using Multistage Training with Transmission Learning

Object localization is a computer vision technique to identify real-world objects such as birds, cats, flowers, cars in images or videos. The algorithm is based on a feature extraction and learning algorithm to recognize instances of an object category. Bird's species are the most amazing creat...

Full description

Saved in:
Bibliographic Details
Main Authors: K N, Raja Praveen, Pasumarty, Rohitha
Format: Conference Proceeding
Language:English
Subjects:
Bird Species
Birds
Computer vision
Data Augmentation
Deep Neural Network
Digital images
Earth
Feature extraction
Google Net
Image Classification Problems
Location awareness
Object Detection
Pre-Trained Model
Training
Transfer Learning
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Object localization is a computer vision technique to identify real-world objects such as birds, cats, flowers, cars in images or videos. The algorithm is based on a feature extraction and learning algorithm to recognize instances of an object category. Bird's species are the most amazing creature exist on earth. They are sensitive to changes in the environment and hence acts as bioindicator species. The main aim of this project is to identify bird species from a high-resolution digital image of Himalayan birds which would help beginner bird watchers or general people for identification. The data sets for the identification of birds are provided by Kaggle which consists of 16 species of birds. For the reduction of the overfitting problem, a data augmentation process is implemented. The model achieves an accuracy of 50.64 or 0.5064% on the dataset of Kaggle.
ISSN:2768-0673
DOI:10.1109/I-SMAC52330.2021.9640676