*Most of the following code include training examples in the ZIP file. Compiled and tested under gcc 4.5.2-8ubuntu4. The code below contains various versions of features and learning algorithms. It is under active development and maintainment. For more details, please contact Zhangzhang Si at zhangzhang.si@gmail.com.

Change Log for hybrid image template code
Date Download Description
2011.5.10
New comparison experiment on INRIA person datset. Code Added a comparison experiment of HIT vs HoG+SVM on INRIA person datset. The data is publicly avaiblable.
2010.8.25
A new dataset introduced in the paper: LHI-ANIMAL-Face in ZIP. The LHI-Animal-Faces dataset is a good dataset for visual learning and classification, because the animal face categories exhibit interesting within-class variation and between-class confusion. The within-class variation includes: (1) rotation and flip transforms, e.g. rotated panda faces and left-or-right oriented pigeon heads; (2) posture variation, e.g. rabbits with standing ears and relaxed ears; and (3) sub-types, e.g. male and female lions. The between-class confusion is mainly caused by shared parts: wolves and cats both have sharp ears at roughly the same positions; sheep and cow faces share a similar contour.
2010.8.20
Updated the learning code for hybrid image templates. Code and Data Added new categories: pear, tomato, coast sunset, highway, mountain, palm tree, forest, frontal car.
Added several display functions to reproduce the figures in the paper.
2010.6.02
Evaluating mixed templates on LHI-AnimalFaces: Code Data It applies the newest version of active basis and mixed template code to learn image templates and perform classification on LHI-AnimalFaces dataset. The package does not include the source images. Please download the LHI-AnimalFaces image set and place them in the "Image/" folder under root directory (the directory that you extract the above package to). To run the baseline algorithm, simply go to the root directory, and type in two commands: global_config; runAll.

The code contains various versions of features and learning algorithms. It is under active development and maintainment. For more details, please contact Zhangzhang Si at zhangzhang.si@gmail.com.
2010.3.10
Evaluating mixed templates on caltech-101: Code First upload. It applies the newest version of active basis and mixed template code to learn image templates and perform classification on Caltech-101 dataset. The package does not include the source images. Please download the Caltech-101 image set, and the Natual image folder and place them in the "Image/" folder under root directory (the directory that you extract the above package to). To run the baseline algorithm, simply go to the root directory, and type in two commands: global_config; runAll.

The code contains various versions of features and learning algorithms. It is under active development and maintainment. For more details, please contact Zhangzhang Si at zhangzhang.si@gmail.com.
2010.2.10
Learning mixed templates: Code (1) Added two more examples: wolf and cow.
(2) Added ZIPProject.m for automatic archiving the code.
2010.1.5
Learning mixed templates: Code Added examples to the learning demo: rabbit head (2 types), fish, motorcycle, hand fan, windsor chair, soccer ball, teapot, elephant head, eagle head, monkey head, dog head, panda head.
2010.1.2
Learning mixed templates: Code Polished learning code and added two templates in the learning demo: palm tree and zebra.
2009
Learning mixed templates: Code First upload.