Learning to count with regression forest and structured labels

By Luca Fiaschi, Ullrich Köthe, Rahul Nair, Fred A. Hamprecht. 2012

232 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

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

In Computer Vision and Pattern Recognition

Citations in the corpus

DeepSetNet: Predicting Sets with Deep Neural Networks (2016, background): “It is important to bear in mind that unlike many existing approaches that learn to count [1, 5, 12, 20, 24, 35, 50, 51], our main goal is not object counting.”
Learning to Count Leaves in Rosette Plants (2015, methodology): “In [15] random forest regression is used to estimate density.”
Improving Object Counting with Heatmap Regulation (2018, methodology): “Extensions to this basic idea are provided in [11,12,13,14,15].”
Interactive Object Counting (2014, methodology): “Similarly to [8, 12], we define the ground truth density from the set of user dotannotations P as the sum of delta functions centred on each of the annotation dots:”
Context-Aware Crowd Counting (2018, methodology): “This trend began in [7, 18, 10], using either Gaussian Process or Random Forests regressors.”
Towards Perspective-Free Object Counting with Deep Learning (2016, methodology): “We follow exactly the same experimental setup that is used in [5, 6, 26, 7].”
Annotated Image Datasets of Rosette Plants (2014, methodology): “We also do hope that this dataset and its future versions, would be used from the computer vision community to learn suitable image statistics [19], adapt and test counting algorithms with [20] and without temporal information [21, 22, 23], segmentation algorithms [24, 25, 26, 27, 28], multi-label s”
CSRNet: Dilated Convolutional Neural Networks for Understanding the Highly Congested Scenes (2018, result): “We compare our approach with the previous state-ofthe-art methods [47, 32, 20, 24].”
Learning to Count with CNN Boosting (2016, methodology): “[10] used random forest regression in order to estimate the object density and improve training efficiency.”
Microscopy cell counting and detection with fully convolutional regression networks (2018, methodology): “1 N/A 3 .”
Learning to Detect Partially Overlapping Instances (2013, methodology): “Moreover, we compare to the counting methods [11, 15] and the detection method [3].”
COUNT Forest: CO-Voting Uncertain Number of Targets Using Random Forest for Crowd Density Estimation (2015, methodology): “The built forest consists of only vector labels in its leaf node, and therefore its memory size is much smaller than regression forest models with dense structured labels proposed in [20, 16].”
Counting in the Wild (2016, background): “As a consequence, models emerged which cast the problem as one where image features were mapped into a global object count [2, 3, 7], or local features mapped into pixel-wise object densities [4, 6, 8, 16, 24, 26] which can be integrated into the object count over”
Composition Loss for Counting, Density Map Estimation and Localization in Dense Crowds (2018, methodology): “The distribution of images from [Flickr, Web, Hajj] for the train and test are [1078, 84, 39] and [306, 21, 7], respectively.”
Extremely Overlapping Vehicle Counting (2015, methodology): “The approach in [8] follows this idea too, but, instead of a linear transformation, the model uses a Regression Forest combined with a spatial average of the estimated densities to make smoother predictions.”
Incorporating Side Information by Adaptive Convolution (2017, background): “Most of the recent state-of-the-art object counting algorithms adopt the density estimation approach (Arteta et al. 2014; Fiaschi et al. 2012; Zhang et al. 2015, 2016; Onoro-Rubio and López-Sastre 2016; Sam et al. 2017; Sindagi and Patel 2017).”
Crowd Counting and Density Estimation by Trellis Encoder-Decoder Networks (2019, methodology): “Earlier approaches strove to compute density maps with hand-crafted features [17, 7] and random forest regressions [7, 29, 49].”