Incompatible shapes vgg 16 + ssd

tensorflow.python.framework.errors_impl.InvalidArgumentError:  Incompatible shapes: [1,8732,4] vs. [1,10,0]
     [[node compute/sub_1 (defined at /losses/smooth_l1_loss.py:22) ]] [Op:__inference_train_function_6041]

Errors may have originated from an input operation.
Input Source operations connected to node compute/sub_1:
 compute/strided_slice_2 (defined at /losses/ssd_loss.py:43)    
 compute/strided_slice_3 (defined at /losses/ssd_loss.py:44)

When i compile a code with a vgg16 train on one class and a SSD from this github (GitHub - Socret360/object-detection-in-keras), i got a problem on the SSD loss function. The problem come from this two line :

bbox_true = y_true[:, :, -12:-8]
bbox_pred = y_pred[:, :, -12:-8]

So if someone have a nee an idea of the problem it will be very helpful