Implemeting Fully Convolutional DenseNets for semantic segmentation.
Hi, i'm working on implementing semantic segmentation on NCS with the 100-layer tiramisu https://arxiv.org/pdf/1611.09326.pdf. However, I don't see NCS support concatenation layers and transpose convolution layer for deconvolution as found in the official repo https://github.com/movidius/ncsdk/releases. Can someone clarify this for me? Also, is it possible for me to run semantic segmentation on NCS with Raspberry Pi 3 at around 7fps? Thank you.