DeepFungusDet: MobileNetV3 Model in Medical Imaging for Fungal Disease Detection

• Published in: 2024 3rd International Conference for Innovation in Technology (INOCON) pp. 1 - 6
• Main Authors: Singh, Gurpreet, Guleria, Kalpna, Sharma, Shagun
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.03.2024
• Subjects: Adaptation models; convolutional neural network; Deep learning; fungal diseases; Fungi; Machine learning; medical; Medical services; Mobile handsets; MobileNetV3; Real-time systems; Training
• DOI: 10.1109/INOCON60754.2024.10511726

A fungal infection in humans is a pathological state resulting from the infiltration and proliferation of fungi within the body. Microorganisms known as fungi are present in the air, water, soil, and plants. The fungal infection can cause skin to become red and inflamed causing bad oral and genital effects in the body. The article presents a deep learning technique for identifying fungal infections using MobileNetV3, which is a compact and resilient convolutional neural network (CNN). The model is trained on a wide variety of fungal datasets, demonstrating its efficiency and mobility for real-time detection using portable devices. The model can categorize and identify various fungal infections across different conditions using its deep learning capabilities. The findings result in an excellent accuracy and speed of the model in identifying fungal infections, indicating its potential for rapid and accessible detection in healthcare, agriculture, and environmental monitoring. The work investigates the effectiveness of the MobileNetV3 model named DeepFungusDet in identifying fungal infections using a broad dataset containing various fungal infections. This model has been implemented at different numbers of epochs resulting in the highest accuracy identification of 93.14% at epoch 13 and a loss of 0.4494, indicating its promise in recognizing fungal infections. This tool provides a portable option for real-time identification via mobile devices, paving the way for future research and use in the crucial field of fungus identification. The findings represent a major step forward in fungus detection and provide prospects for developing accessible and practical diagnostic tools in the healthcare industry and related fields.