Neuromorphic computing seeks to emulate the efficiency and adaptability of biological neural systems, providing a foundation for processing asynchronous sensory data such as event-based vision. Within this paradigm, spiking neural networks (SNNs) enable temporal and sparse information processing through discrete spike communication. This study investigates how short-term synaptic plasticity (STP) can enhance motion segmentation and object detection by dynamically filtering background activity and emphasizing moving entities including vehicles, pedestrians, and two-wheelers. Using the NEST simulator, various neuron–synapse configurations inspired by cortical circuits are evaluated on the 1 Megapixel Automotive Detection and MVSEC Datasets. The results demonstrate that the Tsodyks–Markram synapse model in its depressing form achieves superior performance, showcasing the potential of STP-driven SNNs for adaptive and efficient perception in event-based vision..