A proposed deep learning approach for indoor cactus variant detection

Abstract

Indoor cactus variant detection usually consists in recognizing and categorizing various types of indoor cacti through the application of image analysis and artificial intelligence. This research work introduces a new class of indoor cactus varieties identification technique based on a specific deep learning model called Convolutional Neural Network (CNN). To gather the data, 1200 images of six types of indoor cactus were obtained, and the dataset was expanded to 2620 images to yield higher accuracy. The preprocessed dataset was then fed to the custom CNN architecture designed for identifying cactus variants from the given backgrounds. It can be compared with the other well-known transfer learning models containing Xception, MobileNetV2, DenseNet201 and InceptionV3. The findings highlighted the fact that the proposed custom CNN has the best accuracy compared to the other transfer learning-based models with the recognition accuracy of 99. 62%. This goes to support the effectiveness of the proposed CNN architecture in the accurate differentiation between such visually similar cactus variants.