Deep-feed: An Internet of things-enabled smart feeding system for pets powered by deep learning

Abstract

Internet of things (IoT) encompasses a variety of connected devices and technologies designed to improve care, monitoring, and management of pets. IoT technology enables voice or app-driven control of these feeders, allowing pet guardians to remotely dispense food to their pets anytime. In this paper, a novel deep-feed network has been proposed that combines Image and sensor data classification. The inputs, such as camera (images) and sensor data, are sent to the preprocessing stages, where images are preprocessed using a Bilateral filter, and the data using preprocessing techniques such as tokenization, lemmatization, etc. The preprocessed images are sent to the neural network, like a convolutional neural network (CNN) for image classification and a bidirectional gated recurrent unit (BiGRU) to predict the dog's behavior. Next, these two networks are fused, and the fuzzy concept identifies whether the dogs are near the food or not in a cage. If the dog is near the food cage, the control unit will allocate the food and water through the water pump in the dog cage. Then the control unit gives the order to fill the food and water pumps and alerts the user to identify the food in a cage via the Blynk application. The accuracy of the suggested method can reach 99.95%, compared to 84.9%, 87.58%, and 93.91% for conventional models like the cat's monitoring and feeding systems via IoT (CMFSVI), petification, and global system for mobile communications/general packet radio service (GSM/GPRS). In comparison to the current approaches, the accuracy of the suggested methodology increased by 16.09%, 13.8%, and 3.75%, for existing models like CMFSVI, petification, and GSM/GPRS, respectively.