In a warehouse or a retail setup: Given a frame of reference, our model indicates the percentage of empty spaces with reference to a shelf full of boxes (either rectangular or square) for new inventory.
Our dataset consists of 221 images that we have collected with each image having the label of its % empty space used for regression. Using the labeling program that we have made, it simplifies the process of labeling through the following features:
- Keeps the box width constant
- Allows us to draw multiple rectangles on the image
- Returns the % area of empty spaces by dividing the total area of rectangles by the area of the shelf.
- Renames the file to its labels for ease of import. (Also ensures that different images with the same label are not overwritten.)