Chapter 10 - CNNs Invariance A function f[x] of an image is invariant to a transform t[x] if f(t(x))=f(x) covariant if f(t(x))=t(f(x)) Convolution Conv Padded Conv Strided Kernel Filters Atrous Convolution Channels Receptive field Downsampling Pooling Transposed Conv 1x1 conv Image Classification Alex Net Relu Vgg Res Net Object Detection YOLO Non Maxima Supression Semantic Segmentation Vgg DeconvNet Extras ImageNet Le Net Inception Depthwise Separable PixelShuffle Xavier Initialization Cutout Unet Mixup Batch Normalization
