receiving image data of an item;

determining an attribute value and a prediction confidence value for the item based on the image data, wherein the attribute value and the prediction confidence value are determined using an attribute machine learning system having one or more multi-headed machine learning models, and wherein each of the one or more multi-headed machine learning models is trained to determine attribute values and corresponding prediction confidence values for a particular type of item;

comparing the prediction confidence value to a threshold, wherein if the prediction confidence value is outside the threshold, then the attribute value is selected for user review; and

providing a user review interface during user review, wherein the user review interface includes a common attribute item viewer panel having one or more images of items determined by the attribute machine learning system to have a common attribute value, wherein all items having the common attribute value are viewable within the common attribute item viewer panel without needing to toggle between other panels or between other user interface elements, wherein the user review interface includes an item viewer panel, wherein when an item in the common attribute item viewer panel is selected, the item viewer panel is overlaid above the common attribute item viewer panel, wherein the item viewer panel has an image viewer interface and a selectable attribute interface, wherein the one or more images of the item are viewable via the image viewer interface, wherein a plurality of selectable attribute values is selectable via the selectable attribute interface, wherein the image viewer interface and the selectable attribute interface are adjacent to each other, and wherein the image viewer interface and the selectable attribute interface are presented simultaneously in the item viewer panel without needing to toggle between other panels or between other user interface elements.