Product Details
Platform for Evaluation of Image Classifiers
Created: 2007
Herout Adam, prof. Ing., Ph.D. (DCGM FIT BUT)
Hradiš Michal, Ing., Ph.D. (DCGM FIT BUT)
Chmelař Petr, Ing. (DIFS FIT BUT)
Juránek Roman, Ing., Ph.D. (DCGM FIT BUT)
Šilhavá Jana, Ing. (DCGM FIT BUT)
Zemčík Pavel, prof. Dr. Ing. (DCGM FIT BUT)
Evaluation, classification, image processing, face database.
During the work on the paper "Platform for Evaluation of Image Classifiers", we have created many software tools which together with some shell and awk scripts which form the classifier evaluation platform.
There is some face data and the MNIST hand written digit database available. You can also create your own datasets using our tools. You can use the annotation program to annotate some regions in images and use dataset generator to create dataset from such annotations. The dataset generator resizes the annotated regions to specified size and can also add some random or fixed affine transformations and Gaussian noise. You can mannage the datasets with our merging and mixing tools.
When the data is prepared, you can use our implementation of AdaBoost (odkaz) or any other algorithm to train classifiers and evaluate the results using evaluation scripts.
- z01 framework.z01 10240 KB
- z02 framework.z02 10240 KB
- z03 framework.z03 10240 KB
- z04 framework.z04 10240 KB
- z05 framework.z05 10240 KB
- z06 framework.z06 10240 KB
- z07 framework.z07 10240 KB
- z08 framework.z08 10240 KB
- z09 framework.z09 10240 KB
- z10 framework.z10 10240 KB
- z11 framework.z11 10240 KB
- z12 framework.z12 10240 KB
- z13 framework.z13 10240 KB
- z14 framework.z14 10240 KB
- z15 framework.z15 10240 KB
- zip framework.zip 475 KB