Research on Optimization Algorithm of Computer Vision Based on Deep Learning in Image Recognition

This project aims to improve the adaptive capability of the image recognition system. A new approach is proposed for the purpose of enhancing the precision and robust performance of image recognition by using convolutional neural networks. Firstly, multi-layer convolutions are employed to increase t...

Full description

Saved in:
Bibliographic Details
Published in:2025 IEEE 5th International Conference on Electronic Technology, Communication and Information (ICETCI) pp. 1337 - 1341
Main Author: Dekun, Xian
Format: Conference Proceeding
Language:English
Published: IEEE 23.05.2025
Subjects:
Accuracy
adaptive optimization
Adaptive systems
CNN
Computer vision
Convolutional neural networks
Deep learning
image enhancement
Image recognition
Noise
Optimization
Real-time systems
Transfer learning
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This project aims to improve the adaptive capability of the image recognition system. A new approach is proposed for the purpose of enhancing the precision and robust performance of image recognition by using convolutional neural networks. Firstly, multi-layer convolutions are employed to increase the capacity of image feature extraction, and an adaptive learning policy is applied to optimize the learning speed and weight. To overcome these issues such as bad picture quality and noise, the paper proposes to use the enhanced technique to enhance the performance of the model. The experiment indicates that the proposed method is more effective than other standard datasets in terms of precision of 10%, calculation efficiency of 15%, robust against noise and poor quality images. This algorithm can not only effectively improve image recognition accuracy, but also adapt to the needs of complex environments and images of different quality, and has strong practical application potential. Nevertheless, facing the challenges of larger data sets and real-time applications, the algorithm still has room for improvement in computing resources and real-time processing capabilities.
DOI:10.1109/ICETCI64844.2025.11084241