RECORD
Maximum-Margin Structured Learning with Deep Networks for 3D Human Pose Estimation
Computer Vision and Pattern Recognition · Submitted 27 Aug 2015
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim We test our framework on the Human3.6m dataset and obtain state-of-the-art results compared to other recent methods.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
This paper focuses on structured-output learning using deep neural networks for 3D human pose estimation from monocular images. Our network takes an image and 3D pose as inputs and outputs a score value, which is high when the image-pose pair matches and low otherwise. The network structure consists of a convolutional neural network for image feature extraction, followed by two sub-networks for transforming the image features and pose into a joint embedding. The score function is then the dot-product between the image and pose embeddings. The image-pose embedding and score function are jointly trained using a maximum-margin cost function. Our proposed framework can be interpreted as a special form of structured support vector machines where the joint feature space is discriminatively learned using deep neural networks. We test our framework on the Human3.6m dataset and obtain state-of-the-art results compared to other recent methods. Finally, we present visualizations of the image-pose embedding space, demonstrating the network has learned a high-level embedding of body-orientation and pose-configuration.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | This paper focuses on structured-output learning using deep neural networks for 3D human pose estimation from monocular images. | contribution 0.66 | 0.60 | 0.05 | asserted 0.97 |
| s1 | Our network takes an image and 3D pose as inputs and outputs a score value, which is high when the image-pose pair matches and low otherwise. | method 0.99 | 0.86 | 0.04 | asserted 1.00 |
| s2 | The network structure consists of a convolutional neural network for image feature extraction, followed by two sub-networks for transforming the image features and pose into a joint embedding. | method 1.00 | 0.76 | 0.02 | asserted 1.00 |
| s3 | The score function is then the dot-product between the image and pose embeddings. | method 1.00 | 0.75 | 0.03 | asserted 1.00 |
| s4 | The image-pose embedding and score function are jointly trained using a maximum-margin cost function. | method 0.99 | 0.90 | 0.06 | asserted 1.00 |
| s5 | Our proposed framework can be interpreted as a special form of structured support vector machines where the joint feature space is discriminatively learned using deep neural networks. | interpretation 0.99 | 0.53 | 0.12 | asserted 0.88 |
| s6 | We test our framework on the Human3.6m dataset and obtain state-of-the-art results compared to other recent methods. | result 1.00 | 0.97 | 0.58 | asserted 1.00 |
| s7 | Finally, we present visualizations of the image-pose embedding space, demonstrating the network has learned a high-level embedding of body-orientation and pose-configuration. | result 0.66 | 0.97 | 0.16 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
This paper focuses on structured-output learning using deep neural networks for 3D human pose estimation from monocular images. Our network takes an image and 3D pose as inputs and outputs a score value, which is high when the image-pose pair matches and low otherwise. The network structure consists of a convolutional neural network for image feature extraction, followed by two sub-networks for transforming the image features and pose into a joint embedding. The score function is then the dot-product between the image and pose embeddings. The image-pose embedding and score function are jointly trained using a maximum-margin cost function. Our proposed framework can be interpreted as a special form of structured support vector machines where the joint feature space is discriminatively learned using deep neural networks. We test our framework on the Human3.6m dataset and obtain state-of-the-art results compared to other recent methods. Finally, we present visualizations of the image-pose embedding space, demonstrating the network has learned a high-level embedding of body-orientation and pose-configuration.
Human review
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.