Paddy leaf detection plays a crucial role in the early diagnosis and management of diseases affecting rice crops. The objective is to enhance the accuracy and efficiency of leaf detection methods by combining different algorithms and techniques.This paper identifies different paddy leaf diseases such Blast, Bacterial Leaf Blight (BLB), Sheath blight, Brown spot and Tungro. This paper proposes a hybrid algorithm that combines image processing techniques with machine learning algorithms, such asDiscrete Wavelet Transform (DWT), Stationary Wavelet Transform (SWT), Transverse Dyadic Wavelet Transform (TDWT), DnCNN (Denoising Convolutional Neural Network), Particle Swarm Optimization (PSO), TYDWT+DNCNN and PSO+DNCNN to predict paddy leaf diseases. The deep learning model, such as convolutional neural networks (DNCNN), is trained on a large dataset of labelled leaf images to learn discriminative features. The results demonstrate improved leaf detection accuracy compared to traditional approaches. The algorithm effectively handles variations in leaf colour, size, and shape, enabling reliable detection even in complex field conditions.The proposed algorithm exhibits robustness against noise and illumination variations, enabling accurate and efficient detection of diseased paddy leaves.The proposed hybrid algorithms TYDWT+DNCNN algorithm give high accuracy about 99%.

paddy leaf, TYDWT+DNCNN, PSO+DNCNN