Tensor-based approach to accelerate deformable part models
نویسندگان
چکیده
This article provides next step towards solving speed bottleneck of any system that intensively uses convolutions operations (e.g. CNN). Method described in the article is applied on deformable part models (DPM) algorithm. Method described here is based on multidimensional tensors and provides efficient tradeoff between DPM performance and accuracy. Experiments on various databases, including Pascal VOC, show that the proposed method allows decreasing a number of convolutions up to 4.5 times compared with DPM v.5, while maintaining similar accuracy. If insignificant accuracy degradation is allowable, higher computational gain can be achieved. The method consists of filters tensor decomposition and convolutions shortening using the decomposed filter. Mathematical overview of the proposed method as well as simulation results are provided.
منابع مشابه
I-48: Ultrasound Deformable Model for Virtual Surgery Simulation of Oocyte Retrieval in Infertility Programs
Background The use of a medical simulator should enhance the goals of minimally invasive surgery: patient safety, cosmesis, shortening the length of hospital admissions, and reducing cost. Using an innovative approach to the handling of ultrasound images in virtual reality simulation, this article describes a process that employs a hybrid model of deformable models that can be applied in the te...
متن کاملO-39: Ultrasound Deformable Model for Virtual Surgery Simulation of Oocyte Retrieval in Infertility Programs
Background The use of a medical simulator should enhance the goals of minimally invasive surgery: patient safety, cosmesis, shortening the length of hospital admissions, and reducing cost. Using an innovative approach to the handling of ultrasound images in virtual reality simulation, this article describes a process that employs a hybrid model of deformable models that can be applied in the te...
متن کاملAnalysis and Synthesis of Facial Expressions by Feature-Points Tracking and Deformable Model
Face expression recognition is useful for designing new interactive devices offering the possibility of new ways for human to interact with computer systems. In this paper we develop a facial expressions analysis and synthesis system. The analysis part of the system is based on the facial features extracted from facial feature points (FFP) in frontal image sequences. Selected facial feature poi...
متن کاملA Hybrid 3D Colon Segmentation Method Using Modified Geometric Deformable Models
Introduction: Nowadays virtual colonoscopy has become a reliable and efficient method of detecting primary stages of colon cancer such as polyp detection. One of the most important and crucial stages of virtual colonoscopy is colon segmentation because an incorrect segmentation may lead to a misdiagnosis. Materials and Methods: In this work, a hybrid method based on Geometric Deformable Models...
متن کاملMCCD: Multi-core collision detection between deformable models using front-based decomposition
We present a novel parallel algorithm for fast continuous collision detection (CCD) between deformable models using multi-core processors. We use a hierarchical representation to accelerate these queries and present an incremental algorithm that exploits temporal coherence between successive frames. Our formulation distributes the computation among multiple cores by using finegrained front-base...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
- CoRR
دوره abs/1707.03268 شماره
صفحات -
تاریخ انتشار 2017