Color-Based Free-Space Segmentation Using Online Disparity-Supervised Learning

Sanberg, Willem P.; Dubbelman, Gijs

doi:10.1109/itsc.2015.152

Cited by 11 publications

(17 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For online training, we adopt the training strategies as introduced in [24]. In that work, the stereo-disparity signal is analyzed for several frames, and the resulting segmentation labels are ex-ploited to construct a color model of ground and obstacle regions.…”

Section: Online Trainingmentioning

confidence: 99%

“…In the process of generating the histogram, pixels are weighted with their real-world surface to balance image regions nearby and far away from the camera. With the histograms, class posteriors are generated using Bayes rule, which are subsequently applied in the MAP estimation of the color Stixel World [24]. Additional experiments over different color spaces showed that no single space is optimal for all frames [27].…”

Section: Online Trainingmentioning

confidence: 99%

“…With the histograms, class posteriors are generated using Bayes rule, which are subsequently applied in the MAP estimation of the color Stixel World [24]. Additional experiments over different color spaces showed that no single space is optimal for all frames [27]. In other words, their color representation can be potentially improved by adapting it better to the imaging circumstances.…”

Section: Online Trainingmentioning

confidence: 99%

“…We utilize the publicly available data from Sanberg et al [23] [24] as the training set for our offline training of the FCN. Combined, the data consists of 188 frames with manual annotations of free space.…”

Section: Datasetmentioning

confidence: 99%

“…A state-of-the-art algorithm to distinguish free space and obstacles is the Disparity Stixel World [21]. It performs very well under favourable lighting conditions where the stereo estimation works reliably, but the algorithm is shown to have trouble under adverse conditions such as dim or very bright light [22][23] [24]. We will use this algorithm to generate freespace masks and exploit these as weak training labels.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Free-Space Detection with Self-Supervised and Online Trained Fully Convolutional Networks

Sanberg¹,

Dubbleman²,

With³

2017

Self Cite

View full text Add to dashboard Cite

Recently, vision-based Advanced Driver Assist Systems have gained broad interest. In this work, we investigate free-space detection, for which we propose to employ a Fully Convolutional Network (FCN). We show that this FCN can be trained in a selfsupervised manner and achieve similar results compared to training on manually annotated data, thereby reducing the need for large manually annotated training sets. To this end, our selfsupervised training relies on a stereo-vision disparity system, to automatically generate (weak) training labels for the color-based FCN. Additionally, our self-supervised training facilitates online training of the FCN instead of offline. Consequently, given that the applied FCN is relatively small, the free-space analysis becomes highly adaptive to any traffic scene that the vehicle encounters. We have validated our algorithm using publicly available data and on a new challenging benchmark dataset that is released with this paper. Experiments show that the online training boosts performance with 5% when compared to offline training, both for F max and AP.

show abstract

Section: Online Trainingmentioning

confidence: 99%

Section: Online Trainingmentioning

confidence: 99%

Section: Online Trainingmentioning

confidence: 99%

Section: Datasetmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Free-Space Detection with Self-Supervised and Online Trained Fully Convolutional Networks

Sanberg¹,

Dubbleman²,

With³

2017

Self Cite

View full text Add to dashboard Cite

show abstract

Improved Stixel Estimation Based on Transitivity Analysis in Disparity Space

Saleem

Chien

Rezaei

et al. 2017

Computer Analysis of Images and Patterns

View full text Add to dashboard Cite

We present a novel method for stixel construction using a calibrated collinear trinocular vision system. Our method takes three conjugate stereo images at the same time to measure the consistency of disparity values by means of the transitivity error in disparity space. Unlike previous stixel estimation methods that are built based on a single disparity map, our proposed method introduces a multi-map fusion technique to obtain more robust stixel calculations. We apply a polynomial curve fitting approach to detect an accurate road manifold, using the v-disparity space which is built based on a confidence map. Comparing the depth information from the extracted stixels (using stixel maps) with depth measurements obtained from a highly accurate LiDAR range sensor, we evaluate the accuracy of the proposed method. Experimental results indicate a significant improvement of 13.6% in the accuracy of stixel detection compared to conventional binocular vision.

show abstract

Fast Nonparametric Road Disparity Estimation and Gradient Constrained Obstacle Detection for UGV Navigation

Zeng

2018

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

Color-Based Free-Space Segmentation Using Online Disparity-Supervised Learning

Cited by 11 publications

References 17 publications

Free-Space Detection with Self-Supervised and Online Trained Fully Convolutional Networks

Free-Space Detection with Self-Supervised and Online Trained Fully Convolutional Networks

Improved Stixel Estimation Based on Transitivity Analysis in Disparity Space

Fast Nonparametric Road Disparity Estimation and Gradient Constrained Obstacle Detection for UGV Navigation

Contact Info

Product

Resources

About