Accurate and intelligent management of laboratory equipment is essential in modern university environments, yet traditional manual monitoring methods suffer from inefficiency, lack of timeliness, and weak data integration. To address these challenges, this study proposes a complete detection and management framework based on an improved YOLOv4 model integrated with Internet of Things (IoT) technologies. First, a TV-L1 optical flow-based preprocessing method is designed to extract high-information keyframes from video streams. Then, K-means++ clustering and an Efficient Channel Attention (ECA) module are incorporated into the YOLOv4 architecture to optimize anchor box allocation and channel-wise feature emphasis. The improved model demonstrates superior detection accuracy and robustness under varying lighting and occlusion conditions. Furthermore, a multidimensional user evaluation and a temporal consistency test validate its performance in real-world laboratory monitoring, achieving an accuracy of 0.98, IoU of 0.95, and a low ID switch rate of 1.3 per 100 frames. This research provides an integrated and scalable technical approach to smart laboratory management, supporting real-time monitoring, device accountability, and IoT-based deployment scenarios.