Flower Pictures Recognition Based on the Advanced Convolutional Neural Network with Oxford Flowers 102 Dataset

Jiarui Hu

2024

Abstract

This paper established a model and trained it to recognize pictures of 102 types of oxford flowers by using Convolutional Neural Network (CNN) Because enhancing effectiveness and efficiency while reducing labor costs is the main advantage of autonomic flower classification technology. This study employs the Oxford Flowers 102 dataset and performs a series of random transformations and adjustments in data preparation. The model consists of convolutional layers (extracts local features by convolving kernels with input images), pooling layers (reduces resolution and parameters), and fully connected layers (combines and classifies features) are employed. Besides, a sequential model is created using tf.keras. Sequential class. It contains multiple max pooling layers, one global average pooling layer, three fully connected layers with Rectified Linear Unit (ReLU) activation and L2 regularization along with Dropout layers, and four convolutional layers with different numbers of filters. Eventually it achieves about 70% accuracy in recognizing flower pictures. 8 versions of the model are carried out to construct a better one. The further study plans involve continuous learning and adaptation by exploring more advanced technology and parameters to become more proficient in this field.

Paper Citation

in Harvard Style

Hu J. (2024). Flower Pictures Recognition Based on the Advanced Convolutional Neural Network with Oxford Flowers 102 Dataset. In Proceedings of the 1st International Conference on Modern Logistics and Supply Chain Management - Volume 1: MLSCM; ISBN 978-989-758-738-2, SciTePress, pages 336-341. DOI: 10.5220/0013330800004558

in Bibtex Style

@conference{mlscm24,
author={Jiarui Hu},
title={Flower Pictures Recognition Based on the Advanced Convolutional Neural Network with Oxford Flowers 102 Dataset},
booktitle={Proceedings of the 1st International Conference on Modern Logistics and Supply Chain Management - Volume 1: MLSCM},
year={2024},
pages={336-341},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013330800004558},
isbn={978-989-758-738-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Modern Logistics and Supply Chain Management - Volume 1: MLSCM
TI - Flower Pictures Recognition Based on the Advanced Convolutional Neural Network with Oxford Flowers 102 Dataset
SN - 978-989-758-738-2
AU - Hu J.
PY - 2024
SP - 336
EP - 341
DO - 10.5220/0013330800004558
PB - SciTePress