In Press
M. Startsev, I. Agtzidis, and M. Dorr, “Characterising and Automatically Detecting Smooth Pursuit in a Large-Scale Ground-Truth Data Set of Dynamic Natural Scenes,” Journal of Vision, In Press.
(MS and IA contributed equally)
I. Agtzidis, M. Startsev, and M. Dorr, “360-degree Video Gaze Behaviour: A Ground-Truth Data Set and a Classification Algorithm for Eye Movements,” in ACM Multimedia 2019, In Press.
T. Faßhauer, et al., “Visual exploration of emotional faces in schizophrenia using masks from the Japanese Noh theatre,” Neuropsychologia, In Press.
M. Dorr, et al., “Binocular Summation and Suppression of Contrast Sensitivity in Strabismus, Fusion and Amblyopia,” Frontiers in Human Neuroscience, vol. 13, pp. 234, 2019. Publisher's VersionAbstract
Purpose: Amblyopia and strabismus affect 2%–5% of the population and cause a broad range of visual deficits. The response to treatment is generally assessed using visual acuity, which is an insensitive measure of visual function and may, therefore, underestimate binocular vision gains in these patients. On the other hand, the contrast sensitivity function (CSF) generally takes longer to assess than visual acuity, but it is better correlated with improvement in a range of visual tasks and, notably, with improvements in binocular vision. The present study aims to assess monocular and binocular CSFs in amblyopia and strabismus patients.Methods: Both monocular CSFs and the binocular CSF were assessed for subjects with amblyopia (n = 11), strabismus without amblyopia (n = 20), and normally sighted controls (n = 24) using a tablet-based implementation of the quick CSF, which can assess a full CSF in <3 min. Binocular summation was evaluated against a baseline model of simple probability summation.Results: The CSF of amblyopic eyes was impaired at mid-to-high spatial frequencies compared to fellow eyes, strabismic eyes without amblyopia, and control eyes. Binocular contrast summation exceeded probability summation in controls, but not in subjects with amblyopia (with or without strabismus) or strabismus without amblyopia who were able to fuse at the test distance. Binocular summation was less than probability summation in strabismic subjects who were unable to fuse.Conclusions: We conclude that monocular and binocular contrast sensitivity deficits define important characteristics of amblyopia and strabismus that are not captured by visual acuity alone and can be measured efficiently using the quick CSF.
M. Startsev, I. Agtzidis, and M. Dorr, “1D CNN with BLSTM for Automated Classification of Fixations, Saccades, and Smooth Pursuits,” Behavior Research Methods, vol. 51, no. 2, pp. 556–572, 2019.
L. A. Lesmes and M. Dorr, “Active Learning for Visual Acuity Testing,” in Proceedings of the 2nd International Conference on Applications of Intelligent Systems (APPIS'19), 2019, pp. 26:1-26:6. Publisher's Version
(both authors contributed equally)
M. Startsev and M. Dorr, “Classifying autism spectrum disorder based on scanpaths and saliency,” in Proceedings of 2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW), 2019, pp. 633–636.
I. Agtzidis and M. Dorr, “Getting (More) Real: Bringing Eye Movement Classification from Monitor- to HMD-based Experiments,” in Proceedings of the 11th ACM Symposium on Eye Tracking Research & Applications, New York, NY, USA, 2019, pp. 18:1–18:8. Publisher's Version
M. Startsev, S. Göb, and M. Dorr, “A Novel Gaze Event Detection Metric That Is Not Fooled by Gaze-independent Baselines,” in Proceedings of the 11th ACM Symposium on Eye Tracking Research & Applications, New York, NY, USA, 2019, pp. 22:1–22:9. Publisher's Version
J. Silberg, et al., “Free visual exploration of natural movies in schizophrenia,” European Archives of Psychiatry and Clinical Neuroscience, vol. 269, no. 4, pp. 407-418, 2019. Publisher's Version
M. Startsev and M. Dorr, “360-aware Saliency Prediction with Conventional Image Saliency Predictors,” Signal Processing: Image Communication, vol. 69, pp. 43–52, 2018.
M. Startsev, I. Agtzidis, and M. Dorr, “Deep learning vs.\ manual annotation of eye movements,” in Proceedings of the 2018 ACM Symposium on Eye Tracking Research & Applications (ETRA 18), 2018, pp. Article No.\ 101.
I. Agtzidis and M. Dorr, “Eye Movement Labelling in Head Mounted Display Experiments,” in Scandinavian Workshop on Applied Eye Tracking (SWAET), 2018.
M. Startsev and M. Dorr, “Increasing Video Saliency Model Generalizability by Training for Smooth Pursuit Prediction,” in Proc.İnternational Conference on Computer Vision and Pattern Recognition (CVPR) - Workshop, 2018, pp. 2082–2085.
I. Agtzidis, A. Alashqar, R. Judeh, M. Startsev, E. Vig, and M. Dorr, “Eye movement prediction for naturalistic videos using C3D features,” in International Conference on Computer Vision Workshop on Mutual Benefits of Cognitive and Computer Vision, 2017.
M. Dorr, T. Elze, H. Wang, Z. - L. Lu, P. Bex, and L. Lesmes, “New precision metrics for contrast sensitivity testing,” IEEE Journal of Biomedical and Health Informatics, vol. 22, no. 3, pp. 919-925, 2017.
M. Dorr and E. Vig, “Saliency Prediction for Action Recognition,” in Visual Content Indexing and Retrieval with Psycho-Visual Models, J. Benois-Pineau and P. L. Callet, Ed. Springer, Cham, 2017, pp. 103-124.
M. Dorr, L. A. Lesmes, T. Elze, H. Wang, Z. - L. Lu, and P. J. Bex, “Evaluation of the precision of contrast sensitivity function assessment on a tablet device,” Scientific Reports, vol. 7, pp. 46706, 2017. Publisher's VersionAbstract


The contrast sensitivity function (CSF) relates the visibility of a spatial pattern to both its size and contrast, and is therefore a more comprehensive assessment of visual function than acuity, which only determines the smallest resolvable pattern size. Because of the additional dimension of contrast, estimating the CSF can be more time-consuming. Here, we compare two methods for rapid assessment of the CSF that were implemented on a tablet device.  For a single-trial assessment, we asked 63 myopes and 38 emmetropes to tap the peak of a
"sweep grating" on the tablet's touch screen.  For a more precise assessment, subjects performed 50 trials of the quick CSF method in a 10-AFC letter recognition task.  Tests were performed with and without optical correction, and in monocular and binocular conditions; one condition was measured twice to assess repeatability.

Results show that both methods are highly correlated; using both common and novel measures for test-retest repeatability, however, the quick CSF delivers more precision with testing times of under three minutes.  Further analyses show how a population prior can improve convergence rate of the quick CSF, and how the multi-dimensional output of the quick CSF can provide greater precision than scalar outcome measures.


S. Schenk, M. Dreiser, G. Rigoll, and M. Dorr, “GazeEverywhere: Enabling Gaze-only User Interaction on an Unmodified Desktop PC in Everyday Scenarios,” in Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems (CHI '17), 2017, pp. 3034-3044.