Real-Time Human Pose Estimation via Cascaded Neural Networks Embedded with Multi-task Learning
| Autor: | Makiko Ito, Teruo Ishihara, Satoshi Tanabe, Mitsuru Tomono, Ryosuke Yamanaka |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Artificial neural network
Computer science business.industry Multi-task learning Pattern recognition 010103 numerical & computational mathematics Human body 010501 environmental sciences Machine learning computer.software_genre 01 natural sciences Convolutional neural network Running time Multiple Models Body joints Artificial intelligence 0101 mathematics business Pose computer 0105 earth and related environmental sciences |
| Zdroj: | Computer Analysis of Images and Patterns ISBN: 9783319646978 CAIP (2) |
| DOI: | 10.1007/978-3-319-64698-5_21 |
| Popis: | Deep convolutional neural networks (DCNNs) have recently been applied to Human pose estimation (HPE). However, most conventional methods have involved multiple models, and these models have been independently designed and optimized, which has led to sub-optimal performance. In addition, these methods based on multiple DCNNs have been computationally expensive and unsuitable for real-time applications. This paper proposes a novel end-to-end framework implemented with cascaded neural networks. Our proposed framework includes three tasks: (1) detecting regions which include parts of the human body, (2) predicting the coordinates of human body joints in the regions, and (3) finding optimum points as coordinates of human body joints. These three tasks are jointly optimized. Our experimental results demonstrated that our framework improved the accuracy and the running time was 2.57 times faster than conventional methods. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|