iLog 3.0: Estimating Food Volume from 2D Images Using Mask R-CNN and Monocular Depth Estimation

Indira Devi Siripurapu, Alakananda Mitra, Saraju P. Mohanty, Elias Kougianos

Abstract: To maintain a healthy and balanced lifestyle, it is essential to consume food in proportion to one’s individual needs. The proportions of the food consumed need to be calculated to determine the exact calorie intake or to keep a log. The proposed system, iLog 3.0, automatically determines the volume or quantity of the food item when uploaded using a mobile application, utilizing state-of-the-art object detection and depth estimation techniques for 2D RGB images. The food item will be identified using the Mask R-CNN (Mask Region-Based Convolutional Neural Network) technique. To determine the height of the food item, the MiDaS (Mixed Depth and Scale) technique is employed to generate a depth map, from which the height is subsequently determined. A high success rate has been achieved, and quantification is accurate compared to the previously used models.

External IDs:dblp:conf/isvlsi/SiripurapuMMK25