3D Extended Histogram of Oriented Gradients (3DHOG) for Classification of Road Users in Urban Scenes

Buch, Norbert; Orwell, James; Velastín, Sergio A.

doi:10.5244/c.23.15

Cited by 56 publications

(50 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We achieved an overall accuracy as high as 95.85% in case of sedans vs vans vs taxis using PCA + DFVS. In the difficult case of sedans vs taxis, we achieved a 97.57% accuracy using PCA + DFVS which is higher than any published results using this dataset [6,7]. PCA + DIVS yielded 99.25% accuracy in case of cars vs vans.…”

Section: Introductionmentioning

confidence: 47%

“…For the case of sedans vs taxis, we use 50 images from each class for training, while 200 images of sedans and 130 images of taxis are used for testing. We use the same experimental setup as that used in [6,7], so that a fair comparison is performed. In the case of sedans vs vans vs taxis, we use 50 images of each class for training and 200 images of sedans, and 200 images of vans and 130 images of taxis for testing.…”

Section: Methodsmentioning

confidence: 99%

“…However, this approach is global feature based; therefore, it is not best suited for cases with partial occlusion. Recently, Buch et al presented a traffic video surveillance system which employs motion 3D extended histogram of oriented gradients (3DHOG) to classify road users [6].…”

Section: Other Approachesmentioning

confidence: 99%

See 2 more Smart Citations

Vehicle classification framework: a comparative study

2014

View full text Add to dashboard Cite

Video surveillance has significant application prospects such as security, law enforcement, and traffic monitoring. Visual traffic surveillance using computer vision techniques can be non-invasive, cost effective, and automated. Detecting and recognizing the objects in a video is an important part of many video surveillance systems which can help in tracking of the detected objects and gathering important information. In case of traffic video surveillance, vehicle detection and classification is important as it can help in traffic control and gathering of traffic statistics that can be used in intelligent transportation systems. Vehicle classification poses a difficult problem as vehicles have high intra-class variation and relatively low inter-class variation. In this work, we investigate five different object recognition techniques: PCA + DFVS, PCA + DIVS, PCA + SVM, LDA, and constellation-based modeling applied to the problem of vehicle classification. We also compare them with the state-of-the-art techniques in vehicle classification. In case of the PCA-based approaches, we extend face detection using a PCA approach for the problem of vehicle classification to carry out multi-class classification. We also implement constellation model-based approach that uses the dense representation of scale-invariant feature transform (SIFT) features as presented in the work of Ma and Grimson (Edge-based rich representation for vehicle classification. Paper presented at the international conference on computer vision, 2006, pp. 1185-1192) with slight modification. We consider three classes: sedans, vans, and taxis, and record classification accuracy as high as 99.25% in case of cars vs vans and 97.57% in case of sedans vs taxis. We also present a fusion approach that uses both PCA + DFVS and PCA + DIVS and achieves a classification accuracy of 96.42% in case of sedans vs vans vs taxis.

show abstract

Section: Introductionmentioning

confidence: 47%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Vehicle classification framework: a comparative study

2014

View full text Add to dashboard Cite

show abstract

“…This method requires rear view image of a vehicle including the license plate. Buch et al [3] presents a 3D spatial modeling technique using motion silhouettes which is called 3D HOG. He has shown good results using 3D modeling data in comparison to ordinary histograms and FFT.…”

Section: Literature Reviewmentioning

confidence: 99%

Vehicle Classification using Density based Multi-feature Approach in Support Vector Machine Classifier

Mishra

Banerjee²

2013

IJCA

View full text Add to dashboard Cite

A density based multi-feature combination approach to classify vehicles using support vector machine (SVM) classifier is proposed in this paper. Features like Haar, Gradient, RGB and Pyramidal histogram of oriented gradients are extracted from vehicle images and a kernel density approximation is performed. The results are compared with another set of multi-feature combination and effects of using kernel density estimators and various kernels like Gaussian, Polynomial and Histogram intersection in SVM are also studied. It is found that the above mentioned feature combination produces good results in comparison and effects of scaling, orientation and shadowing were properly compensated. A sample set of 9360 images were used with half of them used for training the classifier and the other half for testing. The vehicles were clustered into four categories, namely 2-wheeler, 3-wheeler, light motor vehicle (LMV) and heavy motor vehicles (HMV) and an accuracy greater than 90% has been observed.

show abstract

“…The most commonly used local features are HOG, HOF, MBH [22], SIFT, 3D-HOG [23], 3D-SIFT [24], and SURF. The motion boundary histogram (MBH) feature descriptor records the motion characteristics of the video content by calculating the histogram of the gradient of optical flow, which is able to discount the camera movement.…”

Section: Related Workmentioning

confidence: 99%

Video Classification Based on Spatial Gradient and Optical Flow Descriptors

Bouzerdoum

Phung

2015

2015 International Conference on Digital Image Computing: Techniques and Applications (DICTA)

View full text Add to dashboard Cite

Feature point detection and local feature extraction are the two critical steps in trajectory-based methods for video classification. This paper proposes to detect trajectories by tracking the spatiotemporal feature points in salient regions instead of the entire frame. This strategy significantly reduces noisy feature points in the background region, and leads to lower computational cost and higher discriminative power of the feature set. Two new spatiotemporal descriptors, namely the STOH and RISTOH are proposed to describe the spatiotemporal characteristics of the moving object. The proposed method for feature point detection and local feature extraction is applied for human action recognition. It is evaluated on three video datasets: KTH, YouTube, and Hollywood2. The results show that the proposed method achieves a higher classification rate, even when it uses only half the number of feature points compared to the dense sampling approach. Moreover, features extracted from the curvature of the motion surface are more discriminative than features extracted from the spatial gradient. Abstract-Feature point detection and local feature extraction are the two critical steps in trajectory-based methods for video classification. This paper proposes to detect trajectories by tracking the spatiotemporal feature points in salient regions instead of the entire frame. This strategy significantly reduces noisy feature points in the background region, and leads to lower computational cost and higher discriminative power of the feature set. Two new spatiotemporal descriptors, namely the STOH and RISTOH are proposed to describe the spatiotemporal characteristics of the moving object. The proposed method for feature point detection and local feature extraction is applied for human action recognition. It is evaluated on three video datasets: KTH, YouTube, and Hollywood2. The results show that the proposed method achieves a higher classification rate, even when it uses only half the number of feature points compared to the dense sampling approach. Moreover, features extracted from the curvature of the motion surface are more discriminative than features extracted from the spatial gradient.

show abstract

3D Extended Histogram of Oriented Gradients (3DHOG) for Classification of Road Users in Urban Scenes

Cited by 56 publications

References 24 publications

Vehicle classification framework: a comparative study

Vehicle classification framework: a comparative study

Vehicle Classification using Density based Multi-feature Approach in Support Vector Machine Classifier

Video Classification Based on Spatial Gradient and Optical Flow Descriptors

Contact Info

Product

Resources

About