I suggest to check by per-image basis.
Let’s say you have 2 reports exported from CP:
- one file named “Image.csv” that reports AreaOccupied
- another file “Objects.csv” that report AreaShape_Area
For e.g. “Image.csv” :
Image 1 : AreaOccupied = 1000
Image 2 : AreaOccupied = 1230
Image 1_Object 1 : AreaShape = 300
Image 1_Object 2 : AreaShape = 200
Image 1_Object 3 : AreaShape = 500
Image 2_Object 1 : AreaShape = 30
Image 2_Object 2 : AreaShape = 200
Image 2_Object 3 : AreaShape = 500
Image 2_Object 4 : AreaShape = 500
If you add Object 1,2,3 of Image 1, it should be equal to Image1_AreaOccupied
If you add Object 1,2,3,4 of Image 2, it should be equal to Image2_AreaOccupied
Please try similar approach for you case, then notice in which Image you have the difference? I suspect the discrepancy should only happen in a few images. Then you can try to trace back to see what really happens in that particular image.
For e.g. you may have filtered object 2 in image 1 in any of the steps in the CP pipeline. If so the summed result in “Objects.csv” will be (AreaShape_Object_1 + AreaShape_Object_3); while in “Image.csv” it’s still the sum of occupied area of 3 original objects.