August 2018
Intermediate to advanced
272 pages
7h 2m
English
Consider the image in previous car image as our test image now. The output of prediction vector for each cell would be:

Note that … entries mean that there will be some random values in predicted vector even for cells which don’t have any object. However in cell 8 the predicted values for x, y, h, w will be hopefully nearly accurate.
At final stage we can filter multiple predicted bounding boxes in each cell using Non-maxima suppression algorithm.
Read now
Unlock full access