This research presents a comprehensive system that utilizes computer vision and deep learning techniques to develop the detection of multiple food methods. Despite the incorporation of deep learning techniques, the effectiveness of the existing detection method for different food products is unsatisfactory due to the utilization of ResNet-101 for feature extraction. The features maps of the ResNet-101 exhibit a size reduction or may vanish entirely following the down-sampling process. The ResNet-101 blocks may have been subject to varying degrees of repetition, with certain blocks receiving a restricted number of repeats and others being excessively repeated. There is an ongoing need to develop the rate of detection in the field of food recognition. The procedure under consideration consists of a series of primary steps. The optimization of the ResNet-101 block entails the careful selection of an appropriate number of repetitions. A supplementary convolutional layer is suggested.