Histogram of Oriented Gradients for Human Detection

This is a simple implementation of Histogram of Oriented Gradients for Human Detection, and used SVM as classifier. Please star if you like this implementation.

Rely ons

$ pip install scikit-learn
$ pip install opencv-python

Use

$ python train.py # for training
$ python evaluate.py # evaluate
$ python seq2jpeg.py # transfer the seq videos to jpg images
$ python vbb2xml.py # transfer the vbb annotations to xml files

Pre-Trained Models

There is a pretrained model in model folder, please use pickle to load the model.

>>>import pickle 
>>>pickle.load(filename)

Dataset

1. IRNIA human Dataset(http://pascal.inrialpes.fr/data/human/)

2. Caltech Pedestrian Detection(http://www.vision.caltech.edu/Image_Datasets/CaltechPedestrians/)

   You need to use seq2jpeg.py and vbb2xml.py to preconditioning the data.