ChewNet: Dataset for Invivo and Invitro Beef and Plant-based Burger Patty Boluses

This dataset examines the chewing dynamics of beef and plant-based burger patties using both human (InVivo) and robotic (InVitro) methods, aiming to (1) identify the optimal robotic chewing cycles that mimic human swallowing and (2) extract bolus texture properties via deep learning-based image analysis. For in vivo trials, three healthy male participants provided near-swallow bolus samples, imaged with a 12MP camera and flatbed scanner, followed by Texture Profile Analysis (TPA). In vitro tests used a 3-DOF chewing robot (ChewBot) with adjustable molar trajectories, artificial saliva (10% food weight), and up to 40 chewing cycles with images, force profiles (100ms intervals), and TPA metrics (after every 5 cycles). ChewNet Dataset is organised into InVivo and InVitro folders, which are crucial for food science research (optimising texture in meat analogues), development of artificial mastication systems for food image processing tasks, and analysing in vitro food bolus properties at the swallowing stage.