Git Product home page Git Product logo

Comments (10)

zhaipro avatar zhaipro commented on May 29, 2024 2

让学习后的模型去识别2015年的验证码,相似图过滤后认为有60K~张不同的图片,列出它自信度超过98%的图片,其中约有6K张。

一些有趣的错误

安全帽:
5 1 699
黑板:
17 2 864
海鸥:
32 0 513
开瓶器:
46 5 579

from easy12306.

zhaipro avatar zhaipro commented on May 29, 2024 1

能不能把真假识别器合并进来,让其有回答"我不知道这是什么"的能力。

从我的直觉上来说,我觉得这是至关重要的能力,因为模型是将整体图片区间(包括仅仅是噪音的图)映射到类别列表,然而显然整体空间的大量内容都是未知的。

一步步来,先试着生成英文验证码,让机器回答说:“不知道这是什么词”。

我觉得应该让机器自己学会如何回答"我不知道"。例如我根本不知道这世上还有多少知识是自己不知道的。

from easy12306.

zhaipro avatar zhaipro commented on May 29, 2024 1

https://www.pyimagesearch.com/2018/04/16/keras-and-convolutional-neural-networks-cnns/
尝试为我们的模型减减肥吧。

from easy12306.

zhaipro avatar zhaipro commented on May 29, 2024

能不能让机器仅仅根据无噪音的文字图学习,最终还能识别有噪音的文字图呢?
https://www.youtube.com/watch?v=XnyM3-xtxHs&list=PLJV_el3uVTsOK_ZK5L0Iv_EQoL1JefRL4

from easy12306.

zhaipro avatar zhaipro commented on May 29, 2024

准备绘画热力图

  1. from keras.applications.vgg16 import preprocess_input, decode_predictions
  2. https://github.com/keras-team/keras-applications/blob/master/keras_applications/imagenet_utils.py
  3. tensorflow/models#517

from easy12306.

zhaipro avatar zhaipro commented on May 29, 2024

新一代的识别力度
[0.12332161949947161, 0.9914363235543496]
[0.07605954375682149, 0.9907580125487536]
[0.04957617516017413, 0.9917754790571477]
[0.05738099154772871, 0.9961845005935221]
[0.044536791779119844, 0.9945735119552315]
[0.061249691417495415, 0.995506189587926]

from easy12306.

zhaipro avatar zhaipro commented on May 29, 2024

这样如何?

依旧用哈希算法来查询图片类型,如果查不到相似的再让深度网络来识别。

from easy12306.

zhaipro avatar zhaipro commented on May 29, 2024

让机器根据哈希统计学结论来学习,之后冲洗哈希库,会怎样呢,哈希的识别力度会提升吗?

from easy12306.

zhaipro avatar zhaipro commented on May 29, 2024

能不能让深入学习后的机器开始怀疑训练集中某样本是错误的呢?不仅让机器根据训练集来学习,而且还要超越训练集。

我人工检测了深入学习后的机器在训练集上犯下的错误,总共统计学专家认定的错误有102个,而我认为其中只有4个是真正的错误,其余98个都是统计学的误差导致的。4个真正的错误中,有3个是统计学提供的答案本身也是错误的,而另一个则是犯下了严重的错误,统计学提供的是正确样本。

它把一个篮球当成了挂钟:
59 0 641 50 108

from easy12306.

zhaipro avatar zhaipro commented on May 29, 2024

统计出错,但机器能给出正确答案的例子:
a
这种图片是不是在我国很流行,因此非常容易获得?从而导致统计量上不去…

from easy12306.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.