Ashish Patel 🇮🇳’s Post

𝗗𝗮𝘆-𝟯𝟵𝟮 𝗖𝗼𝗺𝗽𝘂𝘁𝗲𝗿 𝗩𝗶𝘀𝗶𝗼𝗻 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗲𝗿-𝗯𝗮𝘀𝗲𝗱 𝗦𝗔𝗥 𝗜𝗺𝗮𝗴𝗲 𝗗𝗲𝘀𝗽𝗲𝗰𝗸𝗹𝗶𝗻𝗴 𝗯𝘆 𝗝𝗼𝗵𝗻𝘀 𝗛𝗼𝗽𝗸𝗶𝗻𝘀 𝗨𝗻𝗶𝘃𝗲𝗿𝘀𝗶𝘁𝘆 Follow me for a similar post: Ashish Patel  ------------------------------------------------------------------- 𝗜𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗶𝗻𝗴 𝗙𝗮𝗰𝘁𝘀 : 🔸 Paper: 𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗲𝗿-𝗯𝗮𝘀𝗲𝗱 𝗦𝗔𝗥 𝗜𝗺𝗮𝗴𝗲 𝗗𝗲𝘀𝗽𝗲𝗰𝗸𝗹𝗶𝗻𝗴 🔸 This paper is published arxiv2022. 🔸 Proposed a network architecture that encompasses a transformer-based encoder and a convolution-based decoder, for SAR image despeckling. ------------------------------------------------------------------- 𝗜𝗠𝗣𝗢𝗥𝗧𝗔𝗡𝗖𝗘 🔸 Synthetic Aperture Radar (SAR) images are usually degraded by a multiplicative noise known as speckle which makes processing and interpretation of SAR images difficult.  🔸In this paper, we introduce a transformer-based network for SAR image despeckling.  🔸The proposed despeckling network comprises of a transformer-based encoder which allows the network to learn global dependencies between different image regions - aiding in better despeckling.  🔸The network is trained end-to-end with synthetically generated speckled images using a composite loss function.  🔸Experiments show that the proposed method achieves significant improvements over traditional and convolutional neural network-based despeckling methods on both synthetic and real SAR images. #computervision #artificialintelligence #technology