The paper introduces a lightweight convolutional neural network (CNN) model designed for the classification of four types of cassava leaf diseases: cassava mosaic disease (CMD), cassava green mottle (CGM), cassava bacterial blight (CBB), and cassava brown streak leaf disease (CBSD). The model employs depthwise separable convolution operations to reduce parameter count, making it suitable for mobile devices. It utilizes both channel attention and spatial attention mechanisms to highlight disease areas and suppress background noise, enhancing classification accuracy. The proposed model outperforms state-of-the-art models like VGG16, ResNet 50, EfficientNet, MobileNet V1, and MobileNetV2, using 1.1 million fewer parameters than MobileNet V2. The model's effectiveness is demonstrated through experimental results, achieving a test accuracy of 77% on natural scene images, making it suitable for real-world applications. The paper also discusses the motivation behind the model, its contributions, and the methodology used, including the use of inverted residual blocks and modified attention mechanisms.The paper introduces a lightweight convolutional neural network (CNN) model designed for the classification of four types of cassava leaf diseases: cassava mosaic disease (CMD), cassava green mottle (CGM), cassava bacterial blight (CBB), and cassava brown streak leaf disease (CBSD). The model employs depthwise separable convolution operations to reduce parameter count, making it suitable for mobile devices. It utilizes both channel attention and spatial attention mechanisms to highlight disease areas and suppress background noise, enhancing classification accuracy. The proposed model outperforms state-of-the-art models like VGG16, ResNet 50, EfficientNet, MobileNet V1, and MobileNetV2, using 1.1 million fewer parameters than MobileNet V2. The model's effectiveness is demonstrated through experimental results, achieving a test accuracy of 77% on natural scene images, making it suitable for real-world applications. The paper also discusses the motivation behind the model, its contributions, and the methodology used, including the use of inverted residual blocks and modified attention mechanisms.