Question 1: Single choice
mediumAI-checkedSegmentationDeepLab uses atrous (dilated) convolution. A atrous convolution with rate is applied with stride 1. How large is the input area that one output value sees, and how many weights does the filter have (per input and output channel)?