7-14 Pointwise FeedForward 06:50 7-15 decoder 21:53 7-16 transformer(上) 20:01 7-17 transformer(下) 06:08 7-18 trainer脚本编写 12:30 7-19 infer推理函数编写 13:50 7-20 inference和attention map展示(上) 14:16 7-21 inference和attention map展示(下) 12:11 7-22 本章总结...
To solve these problems, we propose the Parametric Efficient Image InPainting Network (PEIPNet) with a one-stage inpainting framework. First, inspired by [28], we adopt depthwise and pointwise convolutions instead of vanilla convolution to minimize the number of model parameters. Second, we employ...