𝗗𝗮𝘆-𝟭𝟮𝟬 Computer Vision Learning Annotating Object Instances with a 𝗣𝗼𝗹𝘆𝗴𝗼𝗻-𝗥𝗡𝗡 by University of Toronto Follow me for similar post : 🇮🇳 Ashish Patel Interesting Facts : 🔸 This is a paper in CVPR2017 with over 169 citations. 🔸 It Outperforms with 52.13 IoU Score Square Box(35.41), Dilation10(46.80), DeepMask(47.19), SharpMask(52.08) ------------------------------------------------------------------- 𝗔𝗺𝗮𝘇𝗶𝗻𝗴 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 : https://lnkd.in/ejfBXnZ official : https://lnkd.in/eVBtrtM Code : https://lnkd.in/ezCzsHx ------------------------------------------------------------------- 𝗜𝗠𝗣𝗢𝗥𝗧𝗔𝗡𝗖𝗘 🔸 Polygon-RNN : At each time step of the RNN-decoder (right), It feed in an image representation using a modified VGG architecture. Our RNN is a two-layer convolutional LSTM with skip-connection from one and two time steps ago. At the output at each time step, It predict the spatial location of the new vertex of the polygon. #computervision #artificialintelligence #deeplearning #data
For previous post visit this github : https://github.com/ashishpatel26/365-Days-Computer-Vision-Learning-Linkedin-Post
lovely article polygon RNN is smart 🇮🇳 Ashish Patel so architecture is key