Skip to main content
No Access

Food ingredient recognition model via image and textual feature extraction and hybrid classification strategy

Published Online:pp 74-90

This research work focuses on food recognition, especially, the identification of the ingredients from food images. Here, the developed model includes two stages namely: 1) feature extraction; 2) classification. Initially, the image features and textual features will be extracted, where image features like SIFT and improved CNN-based deep features, textural features are extracted. Then, the hybrid classifier is used for the identification of food ingredients that combines the models like neural network (NN) as well as long short-term memory (LSTM). In order to make the accurate results, the weights of NN and LSTM are fine-tuned via the Chebyshev map evaluated teamwork optimisation (CME-TWO) algorithm. At the final stage, the primacy of the offered scheme is proven concerning varied metrics.


food ingredients, improved CNN, TF-IDF features, long short-term memory, LSTM, CME-TWO algorithm