Related Experiment Video
Updated: Sep 24, 2025
08:47
Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation
Published on: February 9, 2024
1.6K
Deep Neural Networks for Automatic Flower Species Localization and Recognition.
Touqeer Abbas1, Abdul Razzaq1, Muhammad Azam Zia2
1Department of Computer Science, MNS University of Agriculture, Multan, Pakistan.
Computational Intelligence and Neuroscience
|May 9, 2022
Summary
This study introduces an optimized deep learning model for automatic flower identification, achieving high accuracy in detecting, localizing, and classifying flower objects using Faster-Recurrent Convolutional Neural Network (Faster-RCNN). The model successfully identifies flower names and class classifications, enhancing computer vision applications.
Area of Science:
- Computer Vision
- Machine Learning
- Deep Learning
Background:
- Deep neural networks excel at image pattern recognition, with object detection widely used in various computer vision tasks.
- Automatic flower identification remains challenging due to class similarities and intra-class variations, necessitating high-quality, precisely labeled data for deep learning models.
Purpose of the Study:
- To develop an optimized and generalized deep convolutional neural network for detecting, localizing, and classifying flower objects.
- To improve the accuracy and efficiency of automatic flower identification systems.
Main Methods:
- Utilized Faster-Recurrent Convolutional Neural Network (Faster-RCNN) and Single Short Detector (SSD) for object detection and classification.
- Trained deep learning models (ResNet 50, ResNet 101, Inception V2, Mobile Net V2) on a dataset of 2000 flower images, with 70% for training, 25% for validation, and 5% for testing.
- Employed a transfer learning approach with the Faster-RCNN model.
Main Results:
- The proposed Faster-RCNN model achieved an optimal mean Average Precision (mAP) score of 83.3% with 300 proposals and 91.3% with 100 proposals on ten flower classes.
- Demonstrated the model's capability to accurately identify, locate, and classify flower objects.
Conclusions:
- The optimized Faster-RCNN model effectively addresses the challenges in automatic flower identification.
- The system provides essential details including flower name, class classification, and supports multilabeling techniques, advancing computer vision applications in botany.
