Oil spills have significant adverse effects on the environment, economy, and human health. They pose dangers to both terrestrial and marine ecosystems. Prompt and accurate identification of affected areas is essential for rapid mitigation and cleanup efforts. However, the process of detecting and manually classifying oil spills in various environments is time-consuming and labor-intensive. Conventional methods often fail to consistently distinguish oil spills from other natural events such as seaweed, sun glare, or marine vegetation, leading to inaccurate alerts. Therefore, this study aims to develop a more efficient and accurate system for classifying images of oil spills using Deep Learning. Our proposed system employs a Convolutional Neural Network (CNN) algorithm to swiftly and accurately evaluate and classify large volumes of images. This paper outlines the main stages and approaches of our proposed model before delving into its implementation. After developing the image classification model, we assessed its accuracy and reliability through performance evaluation. We conducted calculations to determine the impact of varying the number of epochs during training and testing. In addition, we compared our model across three groups of 20 epochs each, observing signs of overfitting at the 31st epoch. Using a batch of 16 processed images with their corresponding classifications, our model achieved an accuracy rate of 87.5%, correctly classifying 14 out of 16 images.