Automatic food bio-hazard detection system

This paper presents the design of a convolutional neural network architecture oriented to detection food waste, generate low, medium, or critical-level alarm. An based on four convolution layers is used, for which database 100 samples prepared. The used with different hyperparameters that make up final architecture, after training process. By means confusion matrix analysis, 100% performance obtained, delivers its output fuzzy system that, depending duration time, generates alarm levels associated risk.