Through a comprehensive evaluation of model complexity and number of parameters, it was determined that the overall performance of the proposed model is the best when eight group convolutions are used ...
Results that may be inaccessible to you are currently showing.
Hide inaccessible results
Feedback