实践 Practice

从真实问题出发,把做过的项目写清楚;点击「打开项目」可直达 GitHub 源码。 Projects built from real problems. “Open project” jumps to the GitHub repo.

2018

破解含语序问题的点击验证码Cracking Sequence Click Captchas

针对按语序点选汉字的验证码:用 YOLO 定位汉字、切割分类,再组成点击序列,解决工商系统一类语序点选验证码。For Chinese click-captchas in reading order: YOLO detection, crop & classify characters, then assemble the click sequence.

深度学习Deep Learning计算机视觉Computer VisionYOLOYOLO开源Open Source

2018

2018年全国大学生计算机应用能力大赛之住房月租金预测大数据赛2018 National Collegiate Contest — Housing Rent Prediction

第一次参加数据分析比赛的复盘:特征工程、模型尝试与改进流程,后续在 A 榜冲到第 3 名。A first data contest retrospective: feature work, modeling, and later reaching 3rd on leaderboard A.

数据分析Data Analysis机器学习Machine Learning竞赛ContestPythonPython

2018

基于Scrapy框架的CrawlSpider类爬取当当全网商品信息Scraping Dangdang with Scrapy CrawlSpider

用 CrawlSpider 抓取当当商品与评论,写入 MySQL,并用中间件应对访问频率限制。Used CrawlSpider to collect Dangdang products and reviews into MySQL, with middleware for rate limits.

爬虫Web ScrapingScrapyScrapyPythonPythonMySQLMySQL