视觉任务相关的问题汇总

深度估计任务

manydepth

问题1：AssertionError: size of input tensor and input format are different. tensor shape: (1, 3, 192, 640), input_format: CHW

解决方法：

writer.add_image("consistency_target/{}".format(j), consistency_target, self.step)
改成：
writer.add_images("consistency_target/{}".format(j), consistency_target, self.step)