INRIAPerson-HOG-SVM

Pedestrian detection is realized based on HOG features and SVM and LR models

My first git :-)

使用方法

HOG提取过程略

LIBSVM

模型训练/验证

命令行调用svm.py，先调用"svm -b train"进行训练，然后调用"svm -b test"在验证集上验证结果

行人检测

在detect_svm.py中指定待检测图像路径，img为灰度图，img_ori为原图，手动调alpha和slidesize大小

res = detect(img,img_ori,alpha = [3],slide = 20)

LR

一体式封装在logicregress.py，train_logister内封装有模型训练、验证、行人检测等所有过程，参数意义同上

res = train_logister(img,img_ori,alpha = [3],slide = 10)