Smart cold storage control using fruit identification for energy-efficient refrigeration

Abstract Optimal cold storages of post-harvest fruits are necessary to enhance preservation, improve energy efficiency, reduce manual monitoring and improve food security along with quality especially in agricultural economies like Pakistan. An experiment using object detection and adaptive temperature control is uniquely performed in a humid subtropical climate. A YOLO (You Only Look Once) object detection algorithm detects categories of fruit to adaptively control the set-point temperature of a cold storage refrigeration system. The algorithm differentiates fresh/stale bananas and apples to control compressor power of a 0.1-ton refrigerator. The object detecting system is 92 % accurate in sensing correct fruit. Adaptive temperature control through the compressor saves 30 % energy for bananas and 12 % for apples compared to conventional systems whilst passively regulating humidity and extending shelf-life of the fruit by 50 %. The methodology of this pilot-study can be extended to different foods beneficial for agriculture and transport sectors.