As rice is the major staple meal for billions of people globally, increasing the quality and quantity of paddy production is essential. Detecting paddy diseases and pests in the early stages of growth is vital for successful production. Living and non-living elements can significantly impact the health of plants, leading to various diseases, including fungi, viroids, nematodes, bacteria, viruses, temperature fluctuations, and nutrient deficiencies. This work proposes a model to detect four rice leaf diseases: BrownSpot, Healthy, Hispa, and Leaf blast. Several research has been done in this field on various datasets, but each has some drawbacks. In recent years, Deep Convolutional Neural Network models have gained popularity as highly effective models for image classification in agricultureAgriculture-related issues, including recognizing plant diseases, counting the number of fruits, and crop monitoring, among other things. This research employed several pre-trained models, namely ResNet50, ResNet18, GoogLeNet, and VGG16. Following that, we presented a new ResNetEnsemble model, the combination of, ResNet50 and ResNet18, pre-trained models. The ResNetEnsemble model accurately classified leaf diseases and achieved the highest accuracy of 88.92% with an outstanding F-score of 85%. In contrast, Pre-trained models showed good performance, and the ResNet50 model acquired the best accuracy of 83.50% with an F-score of 83%, among others. Identifying and predicting rice leaf diseases has been a key research focus in the agricultural information field. Furthermore, the proposed study aims to improve the quality of life, advance technology, foster research development, and promote innovation in industries.