Sequential Optimization for Efficient High-Quality Object Proposal Generation

Zhang, Ziming; Liu, Yun; Chen, Xi; Zhu, Yanjun; Cheng, Ming–Ming; Saligrama, Venkatesh; Torr, Philip H. S.

doi:10.1109/tpami.2017.2707492

Cited by 41 publications

(14 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We have demonstrated how to generate a small set (e.g., 1000) of proposals to cover nearly all potential object regions, using very simple BING features and a postprocessing step. It would be interesting to introduce other additional cues to further reduce the number of proposals while maintaining a high detection rate [88,89], and to explore more applications [23-27, 29, 90] using BING and BING-E. To encourage future work, the source code will be kept up-to-date at http://mmcheng. net/bing.…”

Section: Future Workmentioning

confidence: 99%

BING: Binarized normed gradients for objectness estimation at 300fps

et al. 2019

Self Cite

View full text Add to dashboard Cite

Training a generic objectness measure to produce object proposals has recently become of significant interest. We observe that generic objects with well-defined closed boundaries can be detected by looking at the norm of gradients, with a suitable resizing of their corresponding image windows to a small fixed size. Based on this observation and computational reasons, we propose to resize the window to 8 × 8 and use the norm of the gradients as a simple 64D feature to describe it, for explicitly training a generic objectness measure. We further show how the binarized version of this feature, namely binarized normed gradients (BING), can be used for efficient objectness estimation, which requires only a few atomic operations (e.g., add, bitwise shift, etc.). To improve localization quality of the proposals while maintaining efficiency, we propose a novel fast segmentation method and demonstrate its effectiveness for improving BING's localization performance, when used in multithresholding straddling expansion (MTSE) postprocessing. On the challenging PASCAL VOC2007 dataset, using 1000 proposals per image and intersectionover-union threshold of 0.5, our proposal method achieves a 95.6% object detection rate and 78.6% mean average best overlap in less than 0.005 second per image.

show abstract

Section: Future Workmentioning

confidence: 99%

BING: Binarized normed gradients for objectness estimation at 300fps

et al. 2019

Self Cite

View full text Add to dashboard Cite

show abstract

“…The aim of region proposal is to find the interested regions which potentially contain objects with minimized windows, which is important in multi-object detection tasks. Previous works [6,7] have shown that a simple 8 × 8 feature by computing the normed gradients could be adopted as an efficient region proposal. After binarizing the normed gradients feature, BING could achieve efficient objectness estimation.…”

Section: Region Proposal With Bingmentioning

confidence: 99%

“…And MABO v.s. #WIN means the mean average best overlap for the given #WIN proposals [7]. Hence, a larger DR or MABO value means a better proposal quality.…”

Section: Experiments Setupmentioning

confidence: 99%

“…Recently several computationally intensive approaches have shown a better performance [5], however, it is not practical for them running on embedded platform with limited resources and restricted power consumption. On the other hand, binarized normed gradients (BING) method is proposed as an effective approach for region proposal generation [6,7,8,9], which achieves state-of-the-art detection rate. However, even though there have already been several techniques to optimize the implementation of BING algorithm on CPU [6], due to the control-flow processing style of Von Neumann architecture, it is still difficult for BING to run efficiently especially in embedded platform.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Scalable Pipelined Dataflow Accelerator for Object Region Proposals on FPGA Platform

Yang

Dai

et al. 2018

2018 International Conference on Field-Programmable Technology (FPT)

View full text Add to dashboard Cite

Region proposal is critical for object detection while it usually poses a bottleneck in improving the computation efficiency on traditional control-flow architectures. We have observed region proposal tasks are potentially suitable for performing pipelined parallelism by exploiting dataflow driven acceleration. In this paper, a scalable pipelined dataflow accelerator is proposed for efficient region proposals on FPGA platform. The accelerator processes image data by a streaming manner with three sequential stages: resizing, kernel computing and sorting. First, Ping-Pong cache strategy is adopted for rotation loading in resize module to guarantee continuous output streaming. Then, a multiple pipelines architecture with tiered memory is utilized in kernel computing module to complete the main computation tasks. Finally, a bubble-pushing heap sort method is exploited in sorting module to find the top-k largest candidates efficiently. Our design is implemented with high level synthesis on FPGA platforms, and experimental results on VOC2007 datasets show that it could achieve about 3.67X speedups than traditional desktop CPU platform and >250X energy efficiency improvement than embedded ARM platform.

show abstract

“…In this paper, the network topology attributes and social attributes are used to measure the similarity between nodes, and the hierarchical clustering method effectively divides the community [25]. In the process of data transmission, if the mobile device does not have a suitable transmission target, the message will occupy a lot of cache, and the data transmission in the community is likely to wait a long time and cause the delay in transmission [26]. After dividing the community, we need to further establish the weight distribution between nodes and community to reduce the time complexity and overhead cost and construct a set of candidate relay nodes based on the relationship between information forwarders and adjacent nodes.…”

Section: Introductionmentioning

confidence: 99%

Efficient Data Transmission for Community Detection Algorithm Based on Node Similarity in Opportunistic Social Networks

Xiaokaiti

Qian

2021

Complexity

View full text Add to dashboard Cite

With the rapid development of 5G era, the number of messages on the network has increased sharply. The traditional opportunistic networks algorithm has some shortcomings in processing data. Most traditional algorithms divide the nodes into communities and then perform data transmission according to the divided communities. However, these algorithms do not consider enough nodes’ characteristics in the communities’ division, and two positively related nodes may divide into different communities. Therefore, how to accurately divide the community is still a challenging issue. We propose an efficient data transmission strategy for community detection (EDCD) algorithm. When dividing communities, we use mobile edge computing to combine network topology attributes with social attributes. When forwarding the message, we select optimal relay node as transmission according to the coefficients of channels. In the simulation experiment, we analyze the efficiency of the algorithm in four different real datasets. The results show that the algorithm has good performance in terms of delivery ratio and routing overhead.

show abstract

Sequential Optimization for Efficient High-Quality Object Proposal Generation

Cited by 41 publications

References 59 publications

BING: Binarized normed gradients for objectness estimation at 300fps

BING: Binarized normed gradients for objectness estimation at 300fps

A Scalable Pipelined Dataflow Accelerator for Object Region Proposals on FPGA Platform

Efficient Data Transmission for Community Detection Algorithm Based on Node Similarity in Opportunistic Social Networks

Contact Info

Product

Resources

About