deep-person-reid
deep-person-reid copied to clipboard
multi-scale
How is multi-scale understood? Simulate traditional convolution? The feature scale obtained before the AG gate is the same