搜索资源列表
imageRecognition
- 有两组图片，分别为正样本，和负样本图片，利用SVM训练样本图片，然后输入图片，判断这副图片属于正样本还是负样本，一般用于模式识别中-Two sets of pictures, samples were positive, and negative sample images, the use of SVM training sample images, and then enter the picture, to judge this pair of images belong to the p
kmeans_svm_image_segmentation
- svm 图像分割 用kmeans得到两类分割的图，在显示的图像中用鼠标取点得到2*num个坐标位置的二维向量，前num个为正样本，后num各为负样本-image segmentation
zzz099SVM
- 为了对目标进行快速的检测，提出了一种新的基于支持向量机的级联式分类器的构造方法。该级联分类器由若干个线性SVM弱分类器构成，结构简单，分类时间极快。针对级联结构中的每个节点的训练给出了一个新的SVM框架下的二次规划模型，这使得每个节点都有较高的正样本检测率和适当的负样本错检率。-In order to quickly detect the target, a new cascade based on support vector machine classifier constructor. T
ObjectMarker
- 该软件可以用来opencv训练样本，训练正样本和训练负样本-The software can be used opencv training samples
mousecut
- 基于opencv。直接用鼠标点击，来获取多个图像同意位置的像素值，并直接制作成想要的特征。左键点击获取正样本，右键获取负样本，滚轮点击获取第三类样本。-based on opencv,using mouse to get the value of 6 pictures in the same location.
car
- 550个车辆正样本图像，包含车身侧面，没有负样本-Positive sample images of 550 vehicles, including the side of the body, there is no negative samples
cropnegativeimage
- opencv中处理负样本时随机从大尺寸样本图片中截取负样本图片-opencv random intercept negative sample images from the big picture when dealing with a sample size of negative samples
GenerateHardExamples
- opencv 行人检测中用第一次产生的分类器检测负样本，得出hardexample-The first pedestrian detection using opencv generated classifier to detect negative samples drawn hardexample
traindata
- 用于训练人头分类器的正样本和负样本数据库；训练可使用opencv自带的分类器-Positive samples and negative samples database used to train the head classifier training can use the built-in classifier opencv
SVM_Train_Predict_HOG
- 样本训练 以及训练完成后生成的xml文件 正样本1200个 负样本2400个-Xml file sample training and after training is completed generated negative samples positive samples 1200 2400
negphoto
- 用于人脸检测的负样本4000多张，不是MIT人脸库的照片，请大家放心使用，谢谢谢谢。- negative photos for face detection，negative photos for face detectionnegative photos for face detection
MIT_persons_jpg
- MIT行人数据库，共924张行人图片，原来是PPM格式，很多人找不到转换工具，这里已转换成JPG。该数据库只含正面和背面两个视角，无负样本，未区分训练集和测试集。-MIT pedestrian dataset of JPG format, for pedestrian detection application.
picture-clipping
- 将图片随机剪裁成固定大小的图片，适用于机器学习中剪裁负样本-The image randomly cut to a fixed size image, for machine learning tailoring negative samples
hardexample
- 是指利用第一次训练的分类器在负样本原图上进行行人检测时所有检测到的矩形框，这些矩形框区域很明显都是误报，把这些误报的矩形框保存为图片，加入到初始的负样本集合中，重新进行SVM的训练，可显著减少误报。这种方法叫做自举法-The first is the use of trained classifier all detected rectangle detect pedestrians in the negative when the sample picture, the rectangular
opencv_clipper_picture
- 基于OPENCV，用于生成训练用负样本的工具，将输入图像批量按特定大小截取，保存。-Based on OPENCV, used to generate the training with the tool of negative samples, the input image according to the specific batch size interception, save.
proj4
- 使用滑动窗的人脸检测，滑动窗口能够独立地对图片块进行分类，以确定是否属于被检测目标。内容如下： 1)载入正样本训练集（人脸），并将其转化为HoG特征 2)载入负样本训练集（没有人脸的任意场景），也将其转化为HoG特征 3)使用SVM，对分类器进行训练，训练集包括正训练集和负训练集 4)使用训练好的分类器，在不同的尺度上，对测试集进行分类 -Face detection with a sliding window.
testHogVSM
- 将样本分类为正样本，负样本，难以辨别样本，依次标记为1，-1，-1，通过hog计算各类样本特征，以xml形式写入vsm模型。测试样本时，只需要加载xml文件，计算出测试样本的hog特征，通过预测函数即可得出他属于哪一类别。-The samples are classified into positive samples, negative samples, and the samples are difficult to distinguish. The samples are labeled
NPDFaceDetector_Train
- NPD人脸检测训练demo，学习基于NPD特征的DQT，soft cascade结构用来训练和拒绝负样本-NPD face detection training demo, learning-based NPD feature DQT, soft cascade structure for training and rejected negative samples
SampleMakingTool
- 用于目标检测制作正负样本，正样本在选取框的周围采样，负样本在图片其他位置采样。(For target detection, making positive and negative samples, positive samples are sampled around the selection box, and negative samples are taken at other locations in the picture.)
cutNegSamples
- 从一张不含有目标的图像中随机截取一定数量、规定大小的负样本子图像(negative sample images with a specified size are randomly intercepted from an image without a target)