Latent hierarchical structural learning for object detection

By Long Zhu, Yuanhao Chen, Alan L. Yuille, William T. Freeman. 2010

306 citations recorded by the source corpus.

We present a latent hierarchical structural learning method for object detection. An object is represented by a mixture of hierarchical tree models where the nodes represent object parts. The nodes can move spatially to allow both local and global shape deformations. The models can be trained discriminatively using latent structural SVM learning, where the latent variables are the node positions and the mixture component. But current learning methods are slow, due to the large number of parameters and latent variables, and have been restricted to hierarchies with two layers. In this paper we describe an incremental concave-convex procedure (iCCCP) which allows us to learn both two and three layer models efficiently. We show that iCCCP leads to a simple training algorithm which avoids complex multi-stage layer-wise training, careful part selection, and achieves good performance without requiring elaborate initialization. We perform object detection using our learnt models and obtain performance comparable with state-of-the-art methods when evaluated on challenging public PASCAL datasets. We demonstrate the advantages of three layer hierarchies - outperforming Felzenszwalb et al.'s two layer models on all 20 classes.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Detect What You Can: Detecting and Representing Objects Using Holistic Models and Body Parts (2014, methodology): “Some of these methods represent an object with a holistic object (“root”) template attached to a fixed number of parts [9, 26, 4].”
DeepID-Net: multi-stage and deformable deep convolutional neural networks for object detection (2014, methodology): “The features extracted by the AlexNet are then used for learning 200 one-versus-all SVM classifiers for 200 classes.”
Contextualizing Object Detection and Classification (2015, methodology): “[46] using latent hierarchical structural learning, [UCI_2009] by Desai et al.”
From Red Wine to Red Tomato: Composition with Context (2017, background): “The parts-based systems such as Deformable Part Models [15], grammars [25, 44, 59, 62, 67, 69], and AND-OR graphs [54, 61, 66] also rely on compositionality of objects to build recognition systems.”
DeePM: A Deep Part-Based Model for Object Detection and Semantic Part Localization (2015, background): “The region-based convolutional neural network (R-CNN) framework (Girshick et al., 2014; Girshick, 2015; Ren et al., 2015) gave a tremendous performance gain over previous state-of-the-art methods such as the de-formable part models (DPMs) (Felzenszwalb et al., 2010; Zhu et al., 2010).”
Single-Pedestrian Detection Aided by Multi-pedestrian Detection (2013, methodology): “DPM is used [19, 30] to learn contextual cues.”
Latent Hierarchical Model of Temporal Structure for Complex Activity Classification (2014, methodology): “Our model is partially inspired by the work of [40] in which Zhu et al. developed a hierarchical model with deep structure for object detection.”
Exploring the Spatial Hierarchy of Mixture Models for Human Pose Estimation (2012, methodology): “Similarly, [18] uses a set of hierarchical models for each object view for better objection detection.”
Randomized Max-Margin Compositions for Visual Recognition (2014, background): “Moreover, the individual contributions of compositions and their parts are evaluated in separate experiments that demonstrate their potential.”
Object Detection Using Strongly-Supervised Deformable Part Models (2012, unknown): “First, we expect the use of additional part-level supervision to enhance the performance of current DPM detectors [1, 2].”
Segmentation as selective search for object recognition (2011, methodology): “The computation is currently reduced by using a weak classifier with simple-to-compute features [6, 9, 13, 28, 29], and by reducing the number of locations on a coarse grid and with fixed window sizes [6, 9, 27].”
Boosted local structured HOG-LBP for object localization (2011, background): “Detector models mainly include part based models [8] [9] [20] [27] and rigid template models [1] [24] [25].”
Feature representation for statistical-learning-based object detection: A review (2015, background): “The combination of LBP [115] also improves the detection effectiveness.”