ICDAR 2021 公式檢測(cè)冠軍方案(代碼)
向AI轉(zhuǎn)型的程序員都關(guān)注了這個(gè)號(hào)??????
機(jī)器學(xué)習(xí)AI算法工程 公眾號(hào)：datayx
Searching in massive collections of digitized printed scientific documents with queries that are mathematical expressions is a research area scarcely explored. To address this problem, a crucial first step involves the detection of regions that may contain mathematical expressions. This contest aims to tackle this problem and thus, provide several reasons that could be interesting for attracting research groups to participate in this competition:
Groups researching in Mathematical Expression Recognition, at some point, need to address the problem of automatic detection of mathematical expressions in a document;
Participants in this contest will have access to a large labeled dataset;
The method of obtaining labeled data in the IBEM corpus is scalable, so it is expected to increase this collection in the future, and this new data could be used in future editions of this contest.
Method Description
We built our approach on FCOS, A simple and strong anchor-free object detector, with ResNeSt as our backbone, to detect embedded and isolated formulas. We employed ATSS as our sampling strategy instead of random sampling to eliminate the effects of sample imbalance. Moreover, we observed and revealed the influence of different FPN levels on the detection result. Generalized Focal Loss is adopted to our loss. Finally, with a series of useful tricks and model ensembles, our method was ranked 1st in the MFD task.
項(xiàng)目 代碼 獲取方式：
關(guān)注微信公眾號(hào) datayx 然后回復(fù) 公式 即可獲取。
AI項(xiàng)目體驗(yàn)地址 https://loveai.tech
Prerequisites
Linux or macOS (Windows is in experimental support)
Python 3.6+
PyTorch 1.3+
CUDA 9.2+ (If you build PyTorch from source, CUDA 9.0 is also compatible)
GCC 5+
MMCV
This project is based on MMDetection-v2.7.0, mmcv-full>=1.1.5, <1.3 is needed.Note: You need to run pip uninstall mmcv first if you have mmcv installed.If mmcv and mmcv-full are both installed, there will be ModuleNotFoundError.
機(jī)器學(xué)習(xí)算法AI大數(shù)據(jù)技術(shù)
搜索公眾號(hào)添加： datanlp
長(zhǎng)按圖片，識(shí)別二維碼
閱讀過本文的人還看了以下文章：
TensorFlow 2.0深度學(xué)習(xí)案例實(shí)戰(zhàn)
基于40萬表格數(shù)據(jù)集TableBank，用MaskRCNN做表格檢測(cè)
《基于深度學(xué)習(xí)的自然語(yǔ)言處理》中/英PDF
Deep Learning 中文版初版-周志華團(tuán)隊(duì)
【全套視頻課】最全的目標(biāo)檢測(cè)算法系列講解，通俗易懂！
《美團(tuán)機(jī)器學(xué)習(xí)實(shí)踐》_美團(tuán)算法團(tuán)隊(duì).pdf
《深度學(xué)習(xí)入門：基于Python的理論與實(shí)現(xiàn)》高清中文PDF+源碼
python就業(yè)班學(xué)習(xí)視頻，從入門到實(shí)戰(zhàn)項(xiàng)目
2019最新《PyTorch自然語(yǔ)言處理》英、中文版PDF+源碼
《21個(gè)項(xiàng)目玩轉(zhuǎn)深度學(xué)習(xí)：基于TensorFlow的實(shí)踐詳解》完整版PDF+附書代碼
《深度學(xué)習(xí)之pytorch》pdf+附書源碼
PyTorch深度學(xué)習(xí)快速實(shí)戰(zhàn)入門《pytorch-handbook》
【下載】豆瓣評(píng)分8.1,《機(jī)器學(xué)習(xí)實(shí)戰(zhàn):基于Scikit-Learn和TensorFlow》
《Python數(shù)據(jù)分析與挖掘?qū)崙?zhàn)》PDF+完整源碼
汽車行業(yè)完整知識(shí)圖譜項(xiàng)目實(shí)戰(zhàn)視頻(全23課)
李沐大神開源《動(dòng)手學(xué)深度學(xué)習(xí)》，加州伯克利深度學(xué)習(xí)（2019春）教材
筆記、代碼清晰易懂！李航《統(tǒng)計(jì)學(xué)習(xí)方法》最新資源全套！
《神經(jīng)網(wǎng)絡(luò)與深度學(xué)習(xí)》最新2018版中英PDF+源碼
將機(jī)器學(xué)習(xí)模型部署為REST API
FashionAI服裝屬性標(biāo)簽圖像識(shí)別Top1-5方案分享
重要開源！CNN-RNN-CTC 實(shí)現(xiàn)手寫漢字識(shí)別
同樣是機(jī)器學(xué)習(xí)算法工程師，你的面試為什么過不了？
前海征信大數(shù)據(jù)算法：風(fēng)險(xiǎn)概率預(yù)測(cè)
【Keras】完整實(shí)現(xiàn)‘交通標(biāo)志’分類、‘票據(jù)’分類兩個(gè)項(xiàng)目，讓你掌握深度學(xué)習(xí)圖像分類
VGG16遷移學(xué)習(xí)，實(shí)現(xiàn)醫(yī)學(xué)圖像識(shí)別分類工程項(xiàng)目
特征工程(二) :文本數(shù)據(jù)的展開、過濾和分塊
如何利用全新的決策樹集成級(jí)聯(lián)結(jié)構(gòu)gcForest做特征工程并打分？
Machine Learning Yearning 中文翻譯稿
全球AI挑戰(zhàn)-場(chǎng)景分類的比賽源碼(多模型融合)
斯坦福CS230官方指南：CNN、RNN及使用技巧速查（打印收藏）
python+flask搭建CNN在線識(shí)別手寫中文網(wǎng)站
中科院Kaggle全球文本匹配競(jìng)賽華人第1名團(tuán)隊(duì)-深度學(xué)習(xí)與特征工程
不斷更新資源
深度學(xué)習(xí)、機(jī)器學(xué)習(xí)、數(shù)據(jù)分析、python
搜索公眾號(hào)添加： datayx