Coconut maturity identification plays a significant role in postharvest industrial processing and quality control. The manual process of identifying the coconut maturity includes acoustic vibrations, colour, size, and weight features, which require human expertise. The proposed system uses deep learning models to classify the coconut into two classes. In this paper, acoustic vibration information of the coconut, colour, and size are used to identify the coconut maturity levels. The two classes of the coconut classification considered are tender and mature. In determining the maturity of the coconut, three deep learning models used are ShuffleNet, GoogLeNet, and ResNet50. The captured dataset includes both the acoustic vibration information and image data. Both image and acoustic vibration (audio) based classification is performed. The accuracy obtained for ShuffleNet is 75.79%, GoogLeNet is 87.89%, and ResNet50 is 82% when image data is used. The acoustic-based classification is performed using Artificial Neural Network. The accuracy obtained for the audio data is 64%. The limitation of the present model is training and testing is performed on the offline stored images. The future plan is to classify the coconuts in real-time by providing the camera interface to the system.