Device fingerprinting (DFP) is a technique used to identify devices without relying on their network or assigned identities, such as IP address, MAC address, or International Mobile Equipment Identity (IMEI) number. DFP accomplishes this by analyzing data from packets sent by the device over the network. In this study, we focus on DFP using inter-arrival time (IAT) and received power level characteristics. IAT, the interval between two consecutive packets received, is unique for each device due to variations in hardware and software. Previous DFP methods primarily utilized statistical techniques to analyze IAT for device identification. Our work introduces a novel approach that involves plotting IAT and power level characteristics and then processing these graphs for device identification. Leveraging the efficiency of deep learning libraries in image processing, we enhance the accuracy of device identification. We configured four Raspberry Pi devices as IoT devices and used a laptop as a router with an installed packet sniffer application. This application recorded packet data from the connected devices in a log file. To identify the devices, we employed a convolutional neural network (CNN), achieving an accuracy of 85.2%.