<p>Currently, how to develop broadband memristors with high computational efficiency and low power consumption has become a key direction in memristor technology. In this work, we propose a broadband synapse based on Ag/WO_3−<i>x</i>:N/ZnO:N/indium tin oxide (ITO), exhibiting excellent optoelectronic synaptic characteristics. First, the electronic synaptic properties of the device are tested to simulate the plasticity of biological synapses. Next, light of different wavelengths is applied, demonstrating that the device generates a photocurrent response to multiple wavelengths. Notably, when two types of light are applied simultaneously, a composite photocurrent response is obtained from the device. By leveraging the photoresponse characteristics of the device, a single-wavelength and dual-wavelength photocurrent response artificial vision array is built to simulate human visual perception. Additionally, using the electronic synaptic properties of the device, an artificial neural network is constructed and integrated with a Field Programmable Gate Array as the core processor to form a floating-point arithmetic system. It is worth mentioning that the object detection edge computing system developed in this work achieves a 10³-fold reduction in power consumption. This achieves more efficient computing power and lower power consumption for object detection, addressing the issue of insufficient computing power and the inability to perform floating-point arithmetic in embedded neuromorphic computing deployments.</p>