Proper diagnosis of plant diseases is crucial towards enhancing quality and yield of crops within the agricultural systems. In this paper we propose an effective image disease classification on cabbage and its leaves by using Lite Separable Convolutional Neural Network (LS-CNN). The model presented is composed of an 8-layered network that uses depthwise separable convolutions to decreased parameters and computation burden but keeps a high-performance rate. A selected dataset of different kinds of cabbage leaf diseases, such as black rot, downy mildew, and leaf spot, is provided to train and assess the model. The LS-CNN extracts disease-specific patterns effectively on input images and classifies the images with 94% accuracy. The model is highly efficient regarding both memory consumption and inference speed, which makes it applicable to be used in low-resource agriculture environments. The findings confirm that lightweight deep learning models can enable real-time, scalable, and reliable plant disease detection to precision farming.