Ashish Patel 🇮🇳’s Post

Day-28 Computer Vision Learning DilatedNet — Dilated Convolution (Semantic Segmentation) by Princeton University and Intel lab a.k.a. “atrous convolution”, “algorithme à trous” and “hole algorithm” Follow me for similar post : 🇮🇳 Ashish Patel 𝗜𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗶𝗻𝗴 𝗙𝗮𝗰𝘁𝘀 : 🔸 It is published in 2016 #ICLR, which has already got over 4220 citations 🔸 The idea of Dilated Convolution is come from the wavelet decomposition. It is also called “atrous convolution”, “algorithme à trous” and “hole algorithm”. ------------------------------------------------------------------- 𝗔𝗺𝗮𝘇𝗶𝗻𝗴 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 : https://lnkd.in/eA9ySQ4 Official Code : https://lnkd.in/eSAYXSV Tensorflow : https://bit.ly/3a92fuV Keras : https://bit.ly/3a8DYVX Pytorch : https://bit.ly/3t6W03p ------------------------------------------------------------------- 𝗜𝗠𝗣𝗢𝗥𝗧𝗔𝗡𝗖𝗘 🔸 (a) The original 3x3 convolution, 1-dilated, and the receptive field is 3x3; 🔸 (b) Perform 3x3 convolution on the basis of (a), 2-dilated, and the receptive field is 7x7; 🔸 (c) Perform 3x3 convolution on the basis of (b), 4-dilated, and the receptive field is 15x15; #artificialintelligence #computervision #technology #india