A Novel Standalone Internet of Things Enabled Device for Detection of Rotten Fruits in Cold Storage Systems

Detection of rotten fruits plays a vital role in the cold storage environment to guarantee consumer safety and minimize waste. Manual methods are often time-consuming and prone to errors. Also, identification of rotten fruits based on images alone is insufficient as the image detection is only based on color and texture. There may be cases where fruits are rotten from inside without change in external appearance which affects the concentration of ethanol emitted by them. The variations of ethanol concentration can be checked using gas sensor. Hence, an innovative method for automated rotten fruit detection that integrates gas sensor readings and also the image of the rotten fruit using a standalone device using Raspberry Pi is proposed. The proposed system utilizes a camera integrated with Raspberry Pi to collect real-time image of the fruit conditions. Deep Learning algorithms such as VGG16, ResNet50 and DenseNet121 are employed for identifying a suitable model for the proposed system. During the comparative analysis, VGG16 demonstrated the highest accuracy of 99.1%. Hence, VGG16 is chosen for implantation of the proposed system. For further detection, the system is equipped with MQ2 gas sensor for detecting specific gases like ethylene, methane etc. emitted by rotten fruits. Both the sensor data and image data are taken into account for rotten fruit detection. In addition, the system also sends an email alert to the authorized personnel for further decision-making and resource allocation.