4848657

一个较好效果的彩色图像合成代码(分左右两个视窗,很好用(A good effect of color image synthesis code (about two Windows, very useful)

使用bp（反向传播算法）实现分类问题，并观察分类过程中误差的变化。(Use bp (back-propagation algorithm) to achieve the classification and observe the changes in classification process in the error.)

delphi开发的显示字体平滑程序(display font smoothing procedures)

说明：  机器学习专门研究计算机怎样模拟或实现人类的学习行为，以获取新的知识或技 能，为包括人脸识别领域的人工智能的发展做出了极大的贡献。本文探索如何应用机器学习中的一些技术，使计算机更好地完成人脸识别领域中的人脸检测和人脸验证。 在人脸检测方面，针对如何快速、准确地检出人脸的问题，基于DLIB中特征模型，可以快速提取检测出人脸，并且提取出人脸特征点,主要作用就是快速检测定位人脸。在训练过程中，引入了ResNet机器学习算法，该算法采用多层卷积神经网络结构，对人脸进行多层特征提取和描述，得到人脸特征描述符。通过测试可以证实了上述方法能够检测和识别人脸。(In terms of face detection, for the problem of how to detect faces quickly and accurately, based on the feature model in DLIB, you can quickly extract and detect faces, and extract face feature points, the main role is to quickly detect and locate faces. In the training process, the VGG-FACE machine learning algorithm is introduced. This algorithm uses a multi-layer convolutional neural network structure to perform multi-layer feature extraction and description on the face to obtain the face feature descriptor. Tests can confirm that the above method can detect and recognize human faces.)

EM算法。问题背景是，有N条长为L的DNA序列，每条序列里面有一条长尾K的目标序列，但不知道他的位置，目标序列内的核苷酸服从foreground product multinomial distribution，非目标序列服从background multinomial distribution，改程序意在求解两种分布的参数。(EM algorithm. Background to the issue is that there are N pieces of length L of DNA sequences, each sequence which has a target sequence Nagao K, but do not know his position, the target nucleotide sequence within obedience foreground product multinomial distribution, non-target sequence obey background multinomial distribution, the program is intended to change the parameters in solving the two distributions.)

首次适应算法First Fit： 从空闲分区表的第一个表目起查找该表,把最先能够满足要求的空闲区分配给作业,这种方法目()

基于mmdetection框架的深度学习目标检测框架(deep learning object detection based on mmdetection)

说明：  LSCN彩色图像增强算法，PCNN算法的优化改进(algorithm of LSCN color image enhancement)

UTIL MOTION DETECTION AND TRACKING

C++语言下, 关于图论算法的一些模版, 包括一般图最大匹配, km匹配, 最小割等等, 共15个模版(C++ language under the graph theory algorithm on a number of templates, including the maximum matching in general graph, km matching, minimum cut, etc., a total of 15 templates)