|Description (include details on usage, files and paper references)||The Daimler Mono Pedestrian Detection Benchmark dataset contains a large training and test set. The training set contains 15.560 pedestrian samples (image cut-outs at 48x96 resolution) and 6744 additional full images not containing pedestrians for extracting negative samples. The test set contains an independent sequence with more than 21.790 images with 56.492 pedestrian labels (fully visible or partially occluded), captured from a vehicle during a 27 min drive through urban traffic, at VGA resolution (640x480, uncompressed).
As such, the dataset is realistic and about one order of magnitude larger than other datasets at time of publication (8.5 Gb).
specifies two evaluation settings: one "generic" (2D bounding box overlap criterion) and one specific to pedestrian detection onboard a vehicle (3D localization criterion, known ground plane and sensor coverage area provide regions of interest, processing constraints).
It specifies performance metrics both at the frame- and trajectory-level (the latter also allows benchmarking of tracking algorithms).
It provides the baseline performance of three state-of-the-art methods (wavelet-based AdaBoost cascade, HOG/linSVM and a convolutional network NN/LRF) on the specified training and test set.
It is "open": both training and test set are made (freely) available for non-commercial purposes. .
M. Enzweiler and D. M. Gavrila.
Monocular Pedestrian Detection: Survey and Experiments
IEEE Trans. on Pattern Analysis and Machine Intelligence, vol.31, no.12, pp.2179-2195, 2009.