Laboratoire d'Interprétation et de Traitement d'Images et Vidéo (LITIV)

You are here

Contact Us

Phone
(514) 340-4711 #5064

Fax
(514) 340-4658

Email
guillaume-alexandre.bilodeau...

Office
M-3420, Pavillon Mackay et Lassonde

Codes and Datasets

Datasets
PTZ Tracking
Chen et al. PTZ evaluation framework, ICIP 2015

For reproducible evaluation of tracking with a simulated PTZ camera.

PTZ camera simulator and Tracking evaluation framework C++ code and Spherical panoramic videos with annotations for single object tracking evaluation

Download link

Requested Citation Acknowledgment:

Chen, G., St-Charles, P.-L., Bouachir, W., Bilodeau, G.-A., Bergevin, R., Reproducible Evaluation of Pan-Tilt-Zoom Tracking, IEEE International Conference on Image Processing (ICIP 2015), Quebec, QC, Canada, September 27-30, 2015

Thermal-visible registration
LITIV-VAP dataset, ICCV, Multi-Sensor Fusion for Dynamic Scene Understanding workshop, 2017

For Mutual Foreground Segmentation. Augmentation of the VAP dataset with foreground mask and calibration data

Dataset download link

Requested Citation Acknowledgment:

St-Charles P.-L., Bilodeau, G.-A., Bergevin, R., Mutual Foreground Segmentation with Multispectral Stereo Pairs, International Conference on Computer Vision Workshops (ICCV Workshops), Venice, Italy, October 22th-29th, 2017

Bilodeau et al. dataset, Infrared Physics & Technology, 2014

For registering infrared and visible people appearing at different planes.

Sample frames
(Vid2, cut2, 3 persons)
IR VISIBLE
Original rectified frames
Foreground frames

Dataset and ground-truth download link (117 MB)

Requested Citation Acknowledgment:

Bilodeau, G.-A., Torabi, A., St-Charles, P.-L., Riahi, D., Thermal-Visible Registration of Human Silhouettes: a Similarity Measure Performance Evaluation, Infrared Physics & Technology, Vol. 64, May 2014, pp. 79-86

Torabi et al. dataset, CVIU, 2012 / St-Charles et al. CVPRW 2015

For infrared - visible image registration

Important note: St-Charles et al. CVPRW 2015 paper added ground-truth polygons to facilitate evaluation. Ground-truth matrices and videos are the same as the Torabi et al. CVIU 2012 paper.

Sample frames
(Sequence1)
VISIBLE THERMAL IR
Original rectified frames

Dataset and ground-truth download link (100 MB)

Requested Citation Acknowledgment:

Torabi, A., Massé, G., Bilodeau, G.-A, An iterative integrated framework for thermal-visible image registration, sensor fusion, and people tracking for video surveillance applications, Computer Vision and Image Understanding, Vol. 116, Issue 2, 2012, pp. 210-221 

and if you use the polygons:

St-Charles, P.-L., Bilodeau, G.-A., Bergevin, R., Online Multimodal Video Registration Based on Shape Matching, IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (CVPR Workshops), Boston, MA, USA, June 7-12, 2015

Single object tracking
Bouachir et al. dataset, CVIU 2015

For object tracking. Includes occlusion and many distractors. With ground-truth.

  SAMPLE FRAMES DESCRIPTION
wbook The target face is occluded partially by a book several times
wdesk The target face hides partially behind a desk
jp1 The target face of the person with the t-shirt Leonardo is occluded by other faces
jp2 The target face of the person with the t-shirt Leonardo is occluded by other people passing in front

Dataset and ground-truth download link (30 MB)

Requested Citation Acknowledgment:

Bouachir, W., Bilodeau, G.-A., Collaborative part-based tracking using salient predictors, Computer Vision and Image Understanding, Vol. 137, August 2015, pp. 88-101

Source Codes
Multimodal registration
Mutual Foreground Segmentation with Multispectral Stereo Pairs

Simultaneous registration and segmentation of infrared and visible images. Work done by Pierre-Luc St-Charles at LITIV lab. at Polytechnique Montreal. Appeared at ICCVW 2017

[C++ code available here]

If you use this code, please cite: St-Charles P.-L., Bilodeau, G.-A., Bergevin, R.,Mutual Foreground Segmentation with Multispectral Stereo Pairs, International Conference on Computer Vision Workshops (ICCV Workshops), Venice, Italy, October 22th-29th, 2017

Multimodal registration of Videos based on the contour of shapes

It was specifically developped for registration of people silhouettes in infrared and visible images. Work done by Pierre-Luc St-Charles at LITIV lab. at Polytechnique Montreal. Appeared at CVPRW 2015

[C++ code available here]

If you use this code, please cite: St-Charles, P.-L., Bilodeau, G.-A.,Bergevin, R.,Online Multimodal Registration based on Shape Matching, , IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (CVPR Workshops), Boston, MA, USA, June 7-12, 2015

Test of features for infrared/visible silhouette registration

Code to test various feature descriptor for infrared and visible silhouette registration.

[C++ code available here]

If you use this code, please cite: Bilodeau, G.-A., Torabi, A., St-Charles, P.-L., Riahi, D., Thermal-Visible Registration of Human Silhouettes: a Similarity Measure Performance Evaluation, Infrared Physics & Technology, Vol. 64, May 2014, pp. 79-86

Change detection/Background subtraction
PAWCS: Background subtraction with background words with automatic adjustments of local sensitivity

This change detection method, called PAWCS, is based on a non-parametric model where each sample is modeled with LBSP and color information inside background words. Parameters are automatically adjusted based on noise measurements. Work done by Pierre-Luc St-Charles at LITIV lab. at Polytechnique Montreal. Published at IEEE WACV 2015 and IEEE TIP 2016

[C++ code available here]

If you use this code, please cite: St-Charles, P.-L., Bilodeau, G.-A.,Bergevin, R., Universal Background Subtraction Using Word Consensus Models, IEEE Transactions on Image Processing, Vol. 25, Issue 10, 2016, pp. 4768 - 4781

SubSENSE: Background subtraction with Local Binary Similarity Patterns (LBSP) with automatic adjustments of local sensitivity

This change detection method, called SuBSENSE, is based on a non-parametric model where each sample is modeled with LBSP and color information. Parameters are automatically adjusted based on noise measurements. Work done by Pierre-Luc St-Charles at LITIV lab. at Polytechnique Montreal. Published at CVPR 2014 Workshops

[C++ code available here]

If you use this code, please cite: St-Charles, P.-L., Bilodeau, G.-A., Bergevin, R., Flexible Background Subtraction with Self-Balanced Local Sensitivity, in IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (CVPR Workshops), Columbus, OH, USA, June 23-28, 2014

LOBSTER: Background subtraction with Local Binary Similarity Patterns (LBSP)

This change detection method, called LOBSTER, is based on a non-parametric model where each sample is modeled with LBSP and color information. Work done by Pierre-Luc St-Charles at LITIV lab. at Polytechnique Montreal. Published at WACV14

[C++ code available here]

If you use this code, please cite: St-Charles, P.-L., Bilodeau, G.-A., Improving Background Subtraction using Local Binary Similarity Patterns, in IEEE Winter conference on Applications of Computer Vision (WACV14), Steamboat Springs, Colorado, USA, March 24-26, 2014

Tracking
SPiKeS: Superpixel-Keypoints Structure for Robust Visual Tracking

This is the implementation of SPiKES object tracker based on voting, superpixels and keypoints. Work done by François-Xavier Derue at LITIV lab. at Polytechnique Montreal. Published in Machine Vision and Applications

[C++ code available here]

If you use this code, please cite: Derue, F.-X., Bilodeau, G.-A., Bergevin, R., SPiKeS: Superpixel-Keypoints Structure for Robust Visual Tracking, Machine Vision and Applications, 2017

CTSE: Single object tracking with structure encoding

This is the implementation of CTSE Single object tracker based on voting and coherent motion of keypoints. Also use contextual information around the tracked object. Work done by Tanushri Chakravorty at LITIV lab. at Polytechnique Montreal. Published at ICIP 2015

[C++ code available here]

If you use this code, please cite: Chakravorty, T., Bilodeau, G.-A., Granger, E., Contextual Object Tracker with Structure Encoding, Accepted for IEEE International Conference on Image Processing (ICIP 2015), Quebec, QC, Canada, September 27-30, 2015, pp. 4937-4941

Urban tracker: Multiple object tracking for a priori unknown objects

Multiple object tracker based on FREAK keypoints, bounding box interpolation and a state machine to handle occlusions and fragmentation. Work done by Jean-Philippe Jodoin at LITIV lab. at Polytechnique Montreal. Published at WACV14.

[C++ code available here]

If you use this code, please cite: Jodoin, J.-P., Bilodeau, G.-A., Saunier, N., Urban Tracker: Multiple Object Tracking in Urban Mixed Traffic, in IEEE Winter conference on Applications of Computer Vision (WACV14), Steamboat Springs, Colorado, USA, March 24-26, 2014

Activity recognition, object detection
Spatio-Temporal Feedback to Detect and Segment Carried Objects

Carried object detection with spatio-temporal information. Work done by Farnoosh Ghadiri at LITIV Lab. at Polytechnique Montreal. Work was published at BMVC 2017.

[C++ code available here]

If you use this code, please cite: Ghadiri, F., Bergevin, R., Bilodeau, G.-A., Spatio-Temporal Feedback to Detect and Segment Carried Objects, 28th British Machine Vision Conference (BMVC), London, UK, September 4th-7th, 2017

Carried Object Detection based on an Ensemble of Contour Exemplars

Carried object detection based on a contour dictionary. Work done by Farnoosh Ghadiri at LITIV Lab. at Polytechnique Montreal. Work was published at ECCV 2016.

[C++ code available here]

If you use this code, please cite: Ghadiri, F., Bergevin, R., Bilodeau, G.-A., Carried Object Detection based on an Ensemble of Contour Exemplars, 14th European Conference on Computer Vision (ECCV 2016), Amsterdam, the Netherlands, October 8-14, 2016, Part VII, pp. 852-866

Activity recognition with the MoFREAK spatio-temporal descriptor

Action recognition / activity recognition for surveillance scenarios with local binary feature descriptors. Work done by Chris Whiten for the VIVA Research Lab at University of Ottawa and for LITIV Lab. at Polytechnique Montreal. Work was completed for TRECVID 2012, as well as further research in the action recognition domain and published at CRV 2013.

[C++ code available here]

If you use this code, please cite: Whiten, C., Laganiere, R., Bilodeau, G.-A.,Efficient Action Recognition with MoFREAK, in Tenth Conference on Computer and Robot Vision (CRV 2013), Regina, Saskatchewan, Canada, May 28-31, 2013, pp. 319-325