FLOWER IMAGE DETECTION USING TRANSFER LEARNING

Computer Engineering

Image processing is an important part of removing features from photos. However, image processing and image interpreting process in the distribution of low-level statistical statistics is a challenging task. These processes are complex as the captured image contains a lot of noise, and the target objects are touched by light, light. In the case of flower classification, photo editing or pre-processing is an important part of the automatic flower photo recognition system. The classification of a flower image depends on low-level elements, for example, color and texture, in order to define and define the content of the image. In this project, a Convolution Neural Network (CNN)-based approach was proposed in which the pre-trained VGG-16 model was used to extract features from flower images and before the model was trained, further analysis was performed to remove the noise. , improve brightness and improve image quality using the Digital Image Processing (DIP) algorithm. The model is trained with about 2000 images in three classes, and the classification is done by FC, SVM, Naïve Bayes classifier and DT; high accuracy achieved with FC layer and SVM 91.56% and 91.36%.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.