Ashish Patel 🇮🇳’s Post

𝗗𝗮𝘆-𝟭𝟵𝟯 Computer Vision Learning 𝗪𝗲𝗶𝗴𝗵𝘁𝗡𝗲𝘁: Revisiting the Design Space of Weight Networks by 𝗛𝗼𝗻𝗴 𝗞𝗼𝗻𝗴 𝗨𝗻𝗶𝘃𝗲𝗿𝘀𝗶𝘁𝘆 𝗼𝗳 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 𝗮𝗻𝗱 𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝘆 Follow me for similar post :  🇮🇳 Ashish Patel Interesting Facts : 🔸 This is a paper in ECCV2020 with over 9 citations. ------------------------------------------------------------------- 𝗔𝗺𝗮𝘇𝗶𝗻𝗴 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 : https://lnkd.in/eScFwFM Code : https://lnkd.in/eheCg9V ------------------------------------------------------------------- 𝗜𝗠𝗣𝗢𝗥𝗧𝗔𝗡𝗖𝗘 🔸 Presented a conceptually simple, flexible and effective framework for weight generating networks. Approach is general that unifies two current distinct and extremely effective SENet and CondConv into the same framework on weight space. 🔸 The method, called 𝗪𝗲𝗶𝗴𝗵𝘁𝗡𝗲𝘁, generalizes the two methods by simply adding one more grouped fully-connected layer to the attention activation layer. We use the WeightNet, composed entirely of (grouped) fully-connected layers, to directly output the convolutional weight. 🔸WeightNet is easy and memory-conserving to train, on the kernel space instead of the feature space. Because of the flexibility, our method outperforms existing approaches on both ImageNet and COCO detection tasks, achieving better Accuracy-FLOPs and Accuracy-Parameter trade-offs. #computervision #artificialintelligence #data