YOLO-LITE: A Real-Time Object Detection Algorithm Optimized for Non-GPU Computers

Huang, Rachel; Pedoeem, Jonathan; Chen, Cuixian

doi:10.1109/bigdata.2018.8621865

Cited by 436 publications

(177 citation statements)

References 21 publications

Supporting

Mentioning

172

Contrasting

Unclassified

Order By: Relevance

“…Since its emergence, the YOLO 1 algorithm has been one of the fastest and most accurate algorithms for identifying objects. There are three different versions of this algorithm, each one was one of the best in its time and this is approved by different papers (Huang, Pedoeem, & Chen, 2019;J. Redmon & Farhadi, 2016; J. S. D. R. G. A. F. Redmon, 2016;Ren, Fang, & Djahel, 2017;Rijthoven, Swiderska-chadaj, & Ciompi, 2018).…”

Section: Methodsmentioning

confidence: 99%

Urban Vision Development in Order to Monitor Wheelchair Users Based on the Yolo Algorithm

Ahmadi

Argany

Samany‬

et al. 2019

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. Disability has been one of the most important problems of social communities throughout the ages. As population and urbanization have grown dramatically over recent years, this problem has more and more created the gap between people with disabilities and ordinary people in terms of access to resources, social services and social partnerships. Therefore, this study attempts to demonstrate the ratio of presence of wheelchair users in a community compared to the total population of the same community and evaluate their patterns of presence in different conditions, for example, various weather conditions. For this purpose, we used the You Look Only Once version 3 (YOLOv3) algorithm which is a multilayer deep learning object detection tool to analyze and extract wheelchair users from three different sets of images taken by a camera located in an intersection proximate to a rehabilitation center in Quebec, Canada. The results show that the proportion of wheelchair users in the sample community is 7.4%, while the population with mobility disabilities in Canada is 9.6%.

show abstract

Section: Methodsmentioning

confidence: 99%

Urban Vision Development in Order to Monitor Wheelchair Users Based on the Yolo Algorithm

Ahmadi

Argany

Samany‬

et al. 2019

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…Such solutions are computationally expensive, which makes them unfeasible, in practice, to be used in devices with computational constraints, such as memory, computational power, bandwidth and energy [13]. In this regard, some "mobile" CNN architectures have been already proposed [19,22,44,49,66], i.e., lightweight convolutional neural network architectures specifically designed for mobile devices.…”

Section: Related Work a Scene Text Detectionmentioning

confidence: 99%

Pelee-Text++: A Tiny Neural Network for Scene Text Detection

et al. 2020

View full text Add to dashboard Cite

“…Redmon made some small changes to further improve YOLO to become YOLOv3 [RF18], which is said with bigger network but more accurate results. Huang et al proposed YOLO-LITE [HPC18] to support real-time object detection, especially for devices without GPUs, such as a laptop or a cellphone.…”

Section: Related Workmentioning

confidence: 99%

A Multi‐Person Selfie System via Augmented Reality

Jie

Yang

2020

Computer Graphics Forum

View full text Add to dashboard Cite

Figure 1: The process of taking a group photo of two persons without asking for the help from others. We assume at first a female is holding the camera to take a picture of a male. (a) The starting status of capturing the male. Note that toggle button at the lower right corner indicates the status in which we need to wait for the system to turn the detected male into a virtual marker. (b) The ending status of capturing the male. Also note at the same corner the toggle button's status has been changed, indicating the completion of the virtual marker construction, and the readiness for capturing the second person, i.e., the female. Also note that the operation aid tool has been enabled (by clicking on the middle near the left border) so that it can be used to guide the male to take a photo with a similar configuration where the female took the first picture. Here a configuration means the orientation and distance of the camera, with respect to the male when the first picture was taken. (c) the male is holding the camera, and how its configuration differs from the original one is shown on the operation aid tool. (d) The female walks in while the male can keep on adjusting the configuration. (e) Once the male is satisfied with relative positions of the two persons, under the hint of the operation aid tool that the configuration is acceptable, he can take the second picture. (f), the final result of our system. Notice that the smaller images appear near the lower right part of (a)∼(e) show the real scenes observed from the camera.

show abstract

YOLO-LITE: A Real-Time Object Detection Algorithm Optimized for Non-GPU Computers

Cited by 436 publications

References 21 publications

Urban Vision Development in Order to Monitor Wheelchair Users Based on the Yolo Algorithm

Urban Vision Development in Order to Monitor Wheelchair Users Based on the Yolo Algorithm

Pelee-Text++: A Tiny Neural Network for Scene Text Detection

A Multi‐Person Selfie System via Augmented Reality

Contact Info

Product

Resources

About