Average of Synthetic Exact Filters

Bolme, David S.; Draper, Bruce A.; Beveridge, J. Ross

doi:10.1109/cvpr.2009.5206701

Cited by 159 publications

(99 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…More modern approaches such as Average of Synthetic Exact Filters (ASEF) [15] introduce a method of tuning filters for particular tasks. Although ASEF has shown to perform well in eye localization [15] and pedestrian detection [16], a large number of samples are required for training, which makes it too slow for online visual tracking. David S. Bolme et al [10] propose the MOSSE filter, which produces ASEF-like filters from fewer training images.…”

Section: Related Work On Correlation Filter-based Trackersmentioning

confidence: 99%

Robust Scale Adaptive and Real-Time Visual Tracking with Correlation Filters

Pi¹,

Gu³

et al. 2016

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYVisual tracking has been studied for several decades but continues to draw significant attention because of its critical role in many applications. Recent years have seen greater interest in the use of correlation filters in visual tracking systems, owing to their extremely compelling results in different competitions and benchmarks. However, there is still a need to improve the overall tracking capability to counter various tracking issues, including large scale variation, occlusion, and deformation. This paper presents an appealing tracker with robust scale estimation, which can handle the problem of fixed template size in Kernelized Correlation Filter (KCF) tracker with no significant decrease in the speed. We apply the discriminative correlation filter for scale estimation as an independent part after finding the optimal translation based on the KCF tracker. Compared to an exhaustive scale space search scheme, our approach provides improved performance while being computationally efficient. In order to reveal the effectiveness of our approach, we use benchmark sequences annotated with 11 attributes to evaluate how well the tracker handles different attributes. Numerous experiments demonstrate that the proposed algorithm performs favorably against several state-of-the-art algorithms. Appealing results both in accuracy and robustness are also achieved on all 51 benchmark sequences, which proves the efficiency of our tracker.

show abstract

Section: Related Work On Correlation Filter-based Trackersmentioning

confidence: 99%

Robust Scale Adaptive and Real-Time Visual Tracking with Correlation Filters

Pi¹,

Gu³

et al. 2016

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

show abstract

“…The face detection is performed by means of the well-known Viola-Jones [26] approach based frontal face detector. Whenever a face is detected, the system, at first, searches for the eye positions exploiting the Average of Synthetic Exact Filters (ASEF) based detector [3]. Eye positions, if detected, provide a measure to scale and rotate the face candidate to a standard pose.…”

Section: Detection and Registrationmentioning

confidence: 99%

Assessment of deep learning for gender classification on traditional datasets

Coco¹,

Carcagnì²,

Leo³

et al. 2016

2016 13th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS)

View full text Add to dashboard Cite

Deep Learning has becoming a popular and effective way to address a large set of issues. In particular, in computer vision, it has been exploited to get satisfying recognition performance in unconstrained conditions. However, this wild race towards even better performance in extreme conditions has overshadowed an important step i.e. the assessment of the impact of this new methodology on traditional issues on which for years the researchers had worked. This is particularly true for biometrics applications where the evaluation of deep learning has been made directly on newest large and more challencing datasets. This lead to a pure data driven evaluation that makes difficult to analyze the relationships between network configurations, learning process and experienced outcomes. This paper tries to partially fill this gap by applying a DNN for gender recognition on the MORPH dataset and evaluating how a lower cardinality of examples used for learning can bias the recognition performance.

show abstract

“…It is to be noted that the face detector generates false positive-detects a face when there is not one-as well as false negatives-misses a face when it is present in the frame. In order to eliminate the false positives the frames undergo a second detector; this detector is an eye detector [1]. The frames in which no eyes are detected are discarded during the second pass as not containing a face.…”

Section: A Face Image Extractionmentioning

confidence: 99%

“…The alignment of the face region is performed using the eye-center coordinates extracted by the eye detector [1]. The alignment and cropping of the face components are achieved by means of geometric normalization.…”

Section: B Alignment and Croppingmentioning

confidence: 99%

Investigating the Periocular-Based Face Recognition Across Gender Transformation

Mahalingam

Ricanek

Albert

2014

IEEE Trans.Inform.Forensic Secur.

View full text Add to dashboard Cite

This paper introduces a novel face recognition problem domain: the medically altered face for gender transformation. A data set of >1.2 million face images was constructed from wild videos obtained from You Tube of 38 subjects undergoing hormone replacement therapy (HRT) for gender transformation over a period of several months to three years. The HRT achieves gender transformation by severely altering the balance of sex hormones, which causes changes in the physical appearance of the face and body. This paper explores that the impact of face changes due to hormone manipulation and its ability to disguise the face and hence, its ability to effect match rates. Face disguise is achieved organically as hormone manipulation causes pathological changes to the body resulting in a modification of face appearance. This paper analyzes and evaluates face components versus full face algorithms in an attempt to identify regions of the face that are resilient to the HRT process. The experiments reveal that periocular face components using simple texture-based face matchers, local binary patterns, histogram of gradients, and patch-based local binary patterns out performs matching against the full face. Furthermore, the experiments reveal that a fusion of the periocular using one of the simple texture-based approaches (patched-based local binary patterns) out performs two Commercial Off The Shelf Systems full face systems: 1) PittPatt SDK and 2) Cognetic FaceVACs v8.5. The evaluated periocular-fused patch-based face matcher outperforms PittPatt SDK v5.2.2 by 76.83% and Cognetic FaceVACS v8.5 by 56.23% for rank-1 accuracy.

show abstract

Average of Synthetic Exact Filters

Cited by 159 publications

References 15 publications

Robust Scale Adaptive and Real-Time Visual Tracking with Correlation Filters

Robust Scale Adaptive and Real-Time Visual Tracking with Correlation Filters

Assessment of deep learning for gender classification on traditional datasets

Investigating the Periocular-Based Face Recognition Across Gender Transformation

Contact Info

Product

Resources

About