Local Decorrelation For Improved Pedestrian Detection

By Woonhyun Nam, Piotr Doll'ar, J. Han. Neural Information Processing Systems, 2014

389 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/c7056ac425f07984d6fefd4446b4c6028f6405e1

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Yi Yang
2. Xiaodan Liang
3. Liang Zheng
4. Xiaogang Wang
5. Wanli Ouyang
6. Dacheng Tao
7. Jiashi Feng
8. Yunchao Wei

Citations in the corpus

• Learning Complexity-Aware Cascades for Deep Pedestrian Detection (2015, methodology): “Several works proposed alternative feature channels, obtained by convolving different filters with the original HOG+LUV channels [36, 37, 1, 21].”
• Expecting the Unexpected: Training Detectors for Unusual Pedestrians with Adversarial Imposters (2017, unknown): “LDCF: A LDCF detector [33].”
• KrishnaCam: Using a longitudinal, single-person, egocentric dataset for scene understanding tasks (2016, methodology): “Distribution of People Using Dollar’s pedestrian detector [4, 2], we determined that 17% of frames in the dataset contain at least one person (all bars, Figure 4-left).”
• Multispectral Deep Neural Networks for Pedestrian Detection (2016, methodology): “As a canonical case of general object detection problem, in the past decades, pedestrian detection has attracted consistent attention from computer vision community [2, 4, 5, 9, 23, 24, 25, 26, 29, 43, 45].”
• Person Re-identification in the Wild (2016, methodology): “to the effectiveness of proposals for RCNN based methods. These models include Deformable Part Model (DPM) [12], Aggregated Channel Features (ACF) [9] and Locally Decorrelated Channel Features (LDCF) [26]. We also test their respective RCNN versions. We retrain these models on the PRW training set a”
• Illumination-aware Faster R-CNN for Robust Multispectral Pedestrian Detection (2018, methodology): “In these approaches, traditional detectors (usually ICF [21] and its extensions [22, 23, 24]) were used to generate region proposals, followed by a convnet to re-classify the proposed regions.”
• Object detection by labeling superpixels (2015, methodology): “The current publicly available state-of-the-art face detection [37, 5] and pedestrian detection [55, 57] methods are all based on sliding window.”
• Virtual to Real Adaptation of Pedestrian Detectors (2020, background): “The first class of detectors is based on handcrafted features, such as ICF (Integral Channel Features) [14–18].”
• Deep Learning Strong Parts for Pedestrian Detection (2015, methodology): “In the second aspect, we compare the overall result of DeepParts with existing best-performing methods , including VJ [33], HOG [4], MT-DPM [38], MT-DPM+Context [38], JointDeep [23], SDN [17], ACF+SDT [28], InformedHaar [42], ACF-Caltech+ [21], SpatialPooling [27], LDCF [21], AlexNet+ImageNet [14], ”
• Scale-Aware Fast R-CNN for Pedestrian Detection (2015, methodology): “We use dense sampling of the training data (every 4th frame) as adopted in [32], [62] for evaluating all variants of SAF R-CNN.”
• Multispectral pedestrian detection: Benchmark dataset and baseline (2015, methodology): “In addition, the ACF pedestrian detector [9] is widely used as a basis algorithm for the concurrent state-of-the-art pedestrian detectors [20, 22, 29].”
• Is Faster R-CNN Doing Well for Pedestrian Detection? (2016, methodology): “…TA-CNN [4] employs the ACF detector [8] to generate proposals, and trains an R-CNN-style network to jointly optimize pedestrian detection with semantic tasks; the DeepParts method [5] applies the LDCF detector [24] to generate proposals and learns a set of complementary parts by neural networks.”
• Learning Efficient Single-Stage Pedestrian Detectors by Asymptotic Localization Fitting (2018, methodology): “DeepParts [37] uses the LDCF detector [29] to generate proposals and then trains an ensemble of CNN for detecting different parts.”