1:复现faster rcnn代码中,class faster rcnn继承了父类GeneralizedRCNN,开始在写的过程中没使用super.__init()__(),造成了'FasterRCNN' object has no attribute '_modules'
2:python中list乘法真的巨坑,初始值为一个int类型list a = [1, 2, 3],如果 令 a = a * 2 会得到结果 a = [2, 4, 6],如果 令 a = a * 1.2就会报错。can't multiply sequence by non-int of type 'float', 这种算法在pytorch中有所变化,a = torch.as_tensor(a) ,a = a * 1.2会默认为 a = a * 1 向下取整划掉了小数点后面的数
3:matplot,opencv读入图片,height,width,channel = img.shape