A Real-Time FPGA Based Human Detector

Hsiao, Pei‐Yung; Lin, Shih-Yu; Chen, Chuen-Yau

doi:10.1109/is3c.2016.256

Cited by 10 publications

(4 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Compared to the implementation from Negi et al [ 17 ], the memory utilization in this study is significantly lower than theirs, and the resolution is higher. In the implementation from Hsiao et al [ 18 ], their LUTs, registers, DSP blocks, and memory are all less than this study’s. However, their implementation is not accomplished all in hardware.…”

Section: Resultsmentioning

confidence: 66%

“…Some prior works have realized pedestrian detection based on a FPGA, such as [ 16 , 17 , 18 , 19 , 20 , 21 , 22 ]. In early years, Kadota et al [ 16 ] proposed several methods to simplify the computation, such as the square root and arctangent.…”

Section: Introductionmentioning

confidence: 99%

“…Negi et al [ 17 ] proposed an implementation by using binary-patterned HOG features, adaptive boosting (AdaBoost) classifiers [ 23 ], and some approximation arithmetic strategies. Hsiao et al [ 18 ] realized an implementation in an embedded hardware/software co-design. Komorkiewicz et al [ 19 ] proposed an accurate system using single-precision 32-bit floating point computations in all stages of image processing.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Pure FPGA Implementation of an HOG Based Real-Time Pedestrian Detection System

Luo

2018

Sensors

View full text Add to dashboard Cite

In this study, we propose a real-time pedestrian detection system using a FPGA with a digital image sensor. Comparing with some prior works, the proposed implementation realizes both the histogram of oriented gradients (HOG) and the trained support vector machine (SVM) classification on a FPGA. Moreover, the implementation does not use any external memory or processors to assist the implementation. Although the implementation implements both the HOG algorithm and the SVM classification in hardware without using any external memory modules and processors, the proposed implementation’s resource utilization of the FPGA is lower than most of the prior art. The main reasons resulting in the lower resource usage are: (1) simplification in the Getting Bin sub-module; (2) distributed writing and two shift registers in the Cell Histogram Generation sub-module; (3) reuse of each sum of the cell histogram in the Block Histogram Normalization sub-module; and (4) regarding a window of the SVM classification as 105 blocks of the SVM classification. Moreover, compared to Dalal and Triggs’s pure software HOG implementation, the proposed implementation‘s average detection rate is just about 4.05% less, but can achieve a much higher frame rate.

show abstract

Section: Resultsmentioning

confidence: 66%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Pure FPGA Implementation of an HOG Based Real-Time Pedestrian Detection System

Luo

2018

Sensors

View full text Add to dashboard Cite

show abstract

“…The literature also contains reports on HOG implementations in FPGAs, especially for pedestrian detection [18][19][20][21][22][23][24][25][26][27][28][29]. For us, it is interesting to note that they also include some previously introduced simplifications of HOG calculation, which are mostly implementation-related simplifications, rather than intrinsic algorithm changes.…”

Section: Related Workmentioning

confidence: 99%

A New Method of Histogram Computation for Efficient Implementation of the HOG Algorithm

Ilas

2018

Computers

View full text Add to dashboard Cite

In this paper we introduce a new histogram computation method to be used within the histogram of oriented gradients (HOG) algorithm. The new method replaces the arctangent with the slope computation and the classical magnitude allocation based on interpolation with a simpler algorithm. The new method allows a more efficient implementation of HOG in general, and particularly in field-programmable gate arrays (FPGAs), by considerably reducing the area (thus increasing the level of parallelism), while maintaining very close classification accuracy compared to the original algorithm. Thus, the new method is attractive for many applications, including car detection and classification.

show abstract