Abstract: In object detection tasks, small targets are prone to losing critical information during feature extraction by traditional convolutional layers due to their tiny size and sparse features.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback