<![CDATA[The Semantic segmentation models have been used for many things, namely image classification, image detection, and other activities, including outdoor and object segmentation. Those models can either work with having a good result with a custom dataset and give up an excellent accurate process or a bad one. This research aimed to compare two models of semantic segmentation model, namely Unet and Deepvlab, for food images. The research procedure is to create an original food image dataset, process the dataset with two models, analyze the IoU of two models, and compare the mIoU between the models. The research results show that U-net has a higher mIoU value of 0.01 than Deeplab V3 but has less processing time and some parameters. The research results also show that the completeness of performance details and the prediction segmentation results in the Deeplab v3 segmentation model are superior to this research. This research supports previous research findingsregarding the use of U-net and Deeplab v3 in semantic segmentation models. It enriches research on using these models in food image recognition. Further research is needed to evaluate other models in semantic segmentation for food images.]]>
Copyrights © 2023
