A data set providing synthetic and real-world fisheye video sequences

Abstract

In video surveillance as well as automotive applications, so-called fisheye cameras are often employed to capture a very wide angle of view. As such cameras depend on projections quite different from the classical perspective projection, the resulting fisheye image and video data correspondingly exhibits non-rectilinear image characteristics. Typical image and video processing algorithms, however, are not designed for these fisheye characteristics. To be able to develop and evaluate algorithms specifically adapted to fisheye images and videos, a corresponding test data set is therefore introduced in this paper. The first of those sequences were generated during the authors' own work on motion estimation for fisheye videos and further sequences have gradually been added to create a more extensive collection. The data set now comprises synthetically generated fisheye sequences, ranging from simple patterns to more complex scenes, as well as fisheye video sequences captured with an actual fisheye camera. For the synthetic sequences, exact information on the lens employed is available, thus facilitating both verification and evaluation of any adapted algorithms. For the real-world sequences, we provide calibration data as well as the settings used during acquisition. The sequences are freely available via www.lms.lnt.de/fisheyedataset/.

Fisheye Distortion Rectification from Deep Straight Lines

“…In this section, we compare the proposed LaRecNet with the state-of-the-art methods [21], [22], [24] on our proposed SLF dataset, the fisheye video dataset [44] and the images fetched from the Internet. More experimental results can be seen in https://xuezhucun.github.io/LaRecNet.…”

“…For evaluation, we first test our method and the previous state-of-the-art methods on the testing split of the SLF dataset. Then, we test these methods on a public dataset [44] that contains both synthetic and real-world video sequences taken by fisheye cameras. We call the dataset proposed in [44] the fisheye video dataset for the simplicity of representation.…”

“…Then, we test these methods on a public dataset [44] that contains both synthetic and real-world video sequences taken by fisheye cameras. We call the dataset proposed in [44] the fisheye video dataset for the simplicity of representation. Finally, we fetch some fisheye images without available groundtruth from the Internet to qualitatively verify the generalization ability of our proposed method.…”

P‐202: Late‐News Poster: Dynamic Obstacle Detection to Improve BVI Pedestrian's Navigation Decision using CNNs

“…We have been searching for different real-life datasets which are designed for real-life application and not just for scientific experiments. A few datasets have been considered for the application of this study and they are Real-World Fisheye Video Sequences [2], Road Lane Detection (KITTI) [3], and 2d object detection (KITTI) [3]. All the three datasets have the desired urban settings and target obstacles that match our study.…”

The OmniScape Dataset

“…For this reason, they are increasingly used in the field of intelligent vehicles, including fisheye cameras due to their compactness and inexpensive design. Several datasets contain fisheye images, such as CVRG [6], LMS [7], LaFiDa [8], SVMIS [9], "Go Stanford" [10], GM-ATCI [11], and RTH Zurich multi-FoV synthetic datasets [12]. However, it is noted that there is a lack of road scenes omnidirectional images datasets embedded in a vehicle dedicated for computer vision applications.…”