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 ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback