The detection of moving objects in video sequences is a critical task for numerous applications, including autonomous driving, surveillance, and robotics. However, this task becomes significantly more challenging under low-light and adverse weather conditions, where traditional detection methods often fail. This paper presents a novel approach leveraging Convolutional Neural Networks (CNNs) for robust moving object detection in such challenging environments.Our proposed method incorporates advanced CNN architectures specifically designed to handle the complexities introduced by low-light and adverse weather conditions. We integrate a multi-stage preprocessing pipeline that enhances image quality and visibility before feeding the frames into the detection network. Additionally, we employ a temporal convolutional network to effectively utilize temporal information, improving detection accuracy and stability over consecutive frames.Extensive experiments are conducted on benchmark datasets and newly curated video sequences captured under various low-light and adverse weather conditions. The results demonstrate that our approach significantly outperforms state-of-the-art methods in terms of detection accuracy and robustness. Our CNN-based solution not only excels in detecting moving objects but also maintains high performance in real-time applications, proving its practicality and efficiency.This research highlights the potential of advanced CNN techniques in overcoming the limitations posed by challenging environmental conditions, paving the way for more reliable and resilient object detection systems in real-world scenarios.