Artificial satellites and space debris increasingly contaminate astronomical sky images with linear streaks caused by reflected sunlight. These artifacts compromise the quality of long-exposure observations and pose challenges for automated image analysis. These streaks, usually coming from satellites and debris, can also come from near-Earth asteroids (NEAs) that can produce similar patterns when illuminated by the Sun during long exposures. This work explores object detection using Deep Learning to automatically identify these streaks. We train declinations of YOLO and RT-DETR architectures on StreaksYoloDataset, a dataset of 2,388 annotated raw astronomical images captured between 2022 and 2023. The models are evaluated under different background conditions and illumination levels. Our results show that both YOLO models achieve high mean average precision (mAP), demonstrating their effectiveness in detecting streaks even in noisy or low-contrast conditions. This study lays the foundation for automated streak removal or restoration into astronomical image analysis pipelines and opens the door to future integration of generative restoration techniques and intelligent monitoring systems of real-time detection.