Systems, methods, and computer-readable media are disclosed for dynamic nutrition tracking with utensils. Example methods may include receiving an image of a first food item positioned at a first location and a second food item positioned at a second location, the image comprising location metadata, determining, based at least in part on the image and the location metadata, a first set of coordinates corresponding to a first perimeter of the first location of the first food item, and determining a second set of coordinates corresponding to a second perimeter of the second location of the second food item. Example methods may include determining that a utensil performed a gesture associated with a food consumption event, determining that an origination point of the food consumption utensil at a start of the gesture was within the first perimeter, and identifying a weight measurement of a food portion on the utensil.