Poselets: Body part detectors trained using 3D human pose annotations

Bourdev, Lubomir; Malik, Jitendra

doi:10.1109/iccv.2009.5459303

Cited by 827 publications

(761 citation statements)

References 14 publications

Supporting

Mentioning

757

Contrasting

Unclassified

Order By: Relevance

“…Our approach is different from this algorithm in: (1) the type of local descriptors (HOG descriptors on patches vs. boundary fragments or SIFT descriptors on interest points); (2) the usage of alphabet graphemes (strong detectors vs. weak detectors); and (3) the manner of hypothesis verification (string matching vs. Adaboost classifier). Similar to our work, Andriluka et al [1] and Bourdev et al [5,4] learn partbased models to detect people in natural scenes. However, the part prototypes in their algorithms are obtained using detailed annotations of body parts, while in our model the part prototypes are inferred automatically without part annotations.…”

Section: Related Workmentioning

confidence: 63%

“…The learned part prototypes are tightly clustered in both appearance and configuration space (Fig. 2 (b)), which are very much in common with poselets [5,4]. However, different from poselets, which are obtained using manually labeled part regions and keypoints, our part prototypes are automatically learned using human bounding boxes.…”

Section: Part Alphabet Generationmentioning

confidence: 96%

“…Poselets [5,4] are part primitives that are by construction tightly clustered in both appearance and configuration space, for representing and detecting people. At runtime, instances of poselets are found and combined to predict location and extent of humans.…”

Section: Reflectionmentioning

confidence: 99%

“…Human detection in natural images has been an active research topic for decades and has attracted continuous attention from the computer vision community [21,32,1,5,34,41,10]. Though considerable progress has been made in recent years [8,13,20,23], detecting people in uncontrolled environments remains a challenging task.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Human Detection Using Learned Part Alphabet and Pose Dictionary

Yao

Bai

Liu

et al. 2014

Computer Vision – ECCV 2014

View full text Add to dashboard Cite

Abstract. As structured data, human body and text are similar in many aspects. In this paper, we make use of the analogy between human body and text to build a compositional model for human detection in natural scenes. Basic concepts and mature techniques in text recognition are introduced into this model. A discriminative alphabet, each grapheme of which is a mid-level element representing a body part, is automatically learned from bounding box labels. Based on this alphabet, the flexible structure of human body is expressed by means of symbolic sequences, which correspond to various human poses and allow for robust, efficient matching. A pose dictionary is constructed from training examples, which is used to verify hypotheses at runtime. Experiments on standard benchmarks demonstrate that the proposed algorithm achieves state-of-the-art or competitive performance.

show abstract

Section: Related Workmentioning

confidence: 63%

Section: Part Alphabet Generationmentioning

confidence: 96%

Section: Reflectionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Human Detection Using Learned Part Alphabet and Pose Dictionary

Yao

Bai

Liu

et al. 2014

Computer Vision – ECCV 2014

View full text Add to dashboard Cite

show abstract

“…In the image domain, Random Forests have been introduced for human body pose classification [11]. Finally, the combination of holistic and part-based methods has been explored by introducing the concept of Poselets [27] in the pictorial structures framework [2,28]. These approaches have proposed an intermediate representation but they still do not capture the whole anatomy of the human body.…”

Section: Related Workmentioning

confidence: 99%

Holistic Human Pose Estimation with Regression Forests

Belagiannis

Amann

Navab

et al. 2014

Articulated Motion and Deformable Objects

View full text Add to dashboard Cite

Abstract. In this work, we address the problem of human pose estimation in still images by proposing a holistic model for learning the appearance of the human body from image patches. These patches, which are randomly chosen, are used for extracting features and training a regression forest. During training, a mapping between image features and human poses, defined by joint offsets, is learned; while during prediction, the body joints are estimated with an efficient mode-seeking algorithm. In comparison to other holistic approaches, we can recover body poses from occlusion or noisy data. We demonstrate the power of our method in two publicly available datasets and propose a third one. Finally, we achieve state-of-the-art results in comparison to other approaches.

show abstract