Global warming intensifies daily, prompting individuals, NGOs, environmental groups, and governments to combat it through various strategies, such as building va. However, the impact of these efforts on climate change remains understudied. One critical approach involves examining shallow clouds—key to Earth’s radiation balance yet poorly represented in climate models. Changes in low-level cloud frequency and radiative effects suggest global warming may reshape shallow cloud structures. This study analyzes four mesoscale cloud types—Sugar, Flower, Fish, and Gravel—using a method combining SENet feature extraction and U-Net segmentation architecture. State-of-the-art deep learning models, trained on a NASA-curated standard dataset, are fine-tuned for image segmentation, with U-Net serving as the encoder-decoder backbone. The models undergo diverse pre-processing techniques and are assessed via multiple metrics. The top-performing model achieves an F1-Score of 0.663 and a mean IoU score of 0.411.