Getting error : Invalid bounding boxes
13 次查看（过去 30 天）
Hello everyone, I have this kind of problem and i need help.
Error using validateInputData
Bounding box data must be M-by-4 matrices of positive integer values. The following images have invalid bounding box data:
there are so many more images which have invalid boundig boxes... I want to ask : Can i remove these images somehow ? Or how can I solve this problem ?
for example : bounding boxes for this image : E:\ADRIAN\BAKALARKA\DATASET\vsetko2\panasonic_fullhd_01-090-045-202111091200.jpg
   [1407,101.500000000000,179,979] 
I hope someone can help, so many thanks.
As evident from the example you gave, the image has a valid bounding box [1407,101.500000000000,179,979] but the blank brackets “” are responsible for producing the error.
One way would be to remove the empty brackets.
Also make sure that each sample (image) has at-least one bounding box (this means it has at-least one object in the image that can be used for training/testing).
If an image is having multiple bounding boxes, then they should be represented as [bbx1; bbx2; bbx3; …] as represented in the example Object Detection using YOLO v3 Deep Learning. You may try the example with your dataset and see if it still gives empty bracket like “   [1407,101.500000000000,179,979] ”.