It's a machine vision library, written in simple C++ and based on the Torch machine-learning library.
Release 2.1 (070402)
April 2, 2007
There is an important update.
See the ChangeLog.
The new release 2.1 is available.
This release contains few more things such as image warping, morphological operators, Canny edge detection, connected components, 2D Gabor wavelets, image segmentation using the Torch EMTrainer and video support for M2VTS video files (Lum/Cb/Cr 4:2:2 format).
We also simplified the face detection framework and improved it to be able to take into account multi-view faces.
Please still help us. Continue to send us your comments or your suggestions on what is good, what is bad or what is missing (if you want to add a new feature send us your code and we will add it).
If you are using Torch3vision for your scientific research, please refer to this webpage.
- read/write basic image formats (pgm, ppm, gif, tif, jpeg)
- basic image processing and feature extraction algorithms such as rotation, flip, photometric normalisations (Histogram Equalization, Multiscale Retinex, Self-Quotient Image or Gross-Brajovic), edge detection, 2D DCT, 2D FFT, 2D Gabor, PCA to do Eigen-Faces, LDA to do Fisher-Faces
- various metrics (Euclidean, Mahanalobis, ChiSquare, NormalizeCorrelation, TangentDistance, ...)
- face detection using MLP, cascade of Haar-like classifiers
- 2D objects manipulation and drawing
- video file decoding/encoding, image capture from a frame grabber or from USB
- Sony pan/tilt/zoom camera control using VISCA interface
Six reasons to use Torch3vision:
- It is based on Torch3 (so you can use machine learning algorithms with images !!)
- It is free
- It is modular (concept of ipCore)
- It is growing
- It is simple to use
- It contains a lot of nice examples