In the new semester, our Lab, Web Search Mining and Security Lab, plans to hold an academic seminar every Monday, and each time a keynote speaker will share understanding of papers on his\/her related research with you.
<\/p>\n\n\n\n
This week’s seminar is organized as follows: <\/p>\n\n\n\n
Everyone interested in this topic is welcomed to join us. the following is the abstract for this week\u2019s paper.<\/p>\n\n\n\n
\n\t
\n\t\t\tEnd-to-End Text Recognition with Convolutional Neural Networks\n\t\t<\/p>\n\t\t
\n\t\t\tTao Wang David J. Wu Adam\nCoates Andrew Y. Ng\n\t\t<\/p>\n\t\t
\n\t\t\tAbstract\n\t\t<\/p>\n\t\t
\n\t\t\t Full end-to-end text recognition in\nnatural images is a challenging problem that has received much attention\nrecently. Traditional systems in this area have relied on elaborate models\nincorporating carefully hand-engineered features or large amounts of prior\nknowledge. In this paper, we take a different route and combine the\nrepresentational power of large, multilayer neural networks together with\nrecent developments in unsupervised feature learning, which allows us to use a\ncommon framework to train highly-accurate text detector and character\nrecognizer modules. Then, using only simple off-the-shelf methods, we integrate\nthese two modules into a full end-to-end, lexicon-driven, scene text\nrecognition system that achieves state-of-the-art performance on standard\nbenchmarks, namely Street View Text and ICDAR 2003.<\/span>\n\t\t<\/p>\n\t\t \n\t\t\t<\/span>\n\t\t<\/p>\n\t<\/div>\n<\/p>\n\n\n\n Last Monday, Gang Wang<\/strong> gave a presentation about the paper, End-to-End Text Recognition with Convolutional Neural\nNetworks<\/strong>, and shared some opinion on it.<\/p>\n\n\n\n The paper is published on ICPR in 2012. The\nexperiments were taken on two dataset: 1) ICDAR(International Conference on\nDocument Analysis and Recognition ) 2003 Dataset and 2) SVT(Street View Text)\nDataset. So their results were mainly concerned with English.<\/p>\n\n\n\n One question \u201cIn result table2, why I-5 is\nmore accurate than I-50?\u201d was asked. A possible answer is that: 5 and 50 are\nthe number of distractor words provided by other research. The greater the\nnumerical value, the louder the noise.<\/p>\n","protected":false},"excerpt":{"rendered":" NLPIR SEMINAR Y2019#5 INTRO In the new s … \u7ee7\u7eed\u9605\u8bfb NLPIR\nSEMINAR 18th ISSUE COMPLETED<\/strong><\/h2>\n\n\n\n