Usually, the encoder-decoder module consists of an encoder comprising of a CNN which extracts the features from original image
    通常，编码器通常由从原始图像提取特征的CNN组成
    It progressively downsamples the image and reduces feature map resolution to capture high-level details of the image.
    它逐步对图像下采样，并降低feature map的分辨率以捕获图像的高级细节
    The decoder module consists of a set of layers that upsamples the feature map of encoder to recover spatial information
    解码器由一组对编码器的特征图进行上采样以恢复空间信息
    image.png

    part-3_8.pdf