WITHDRAWN: People detection and counting using YOLOv3 and SSD models

Gupta, Pooja; Sharma, Varsha; Varma, Shambhu D.

doi:10.1016/j.matpr.2020.11.562

Cited by 14 publications

(7 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Pertama untuk melatih model yang akan digunakan dan kedua, kumpulan data asli digunakan untuk melakukan deteksi objek. Juga, kita bisa melatih sistemnya juga kita bisa menggunakan sistem yang sudah terlatih [1].…”

Section: Model Terlatihunclassified

Optimasi Mendeteksi Klasifikasi Citra Digital Logo Mobil Indonesia Dengan Metode Single Shot Multibox Detector

Mulyana¹,

Zikri²

2022

Explore. jurnal. sistem. inf. dan. telematika

View full text Add to dashboard Cite

The car logo is one of the features that can identify a vehicle. However, many of the intelligent transportation systems are currently under development and do not yet use a car vehicle recognition system as part of a vehicle identification tool. The previous methods, namely the Local Binary Pattern and Random Forest Methods, had low recognition rates for most small vehicle logos and poor performance under complex environments. The aim of this research is to introduce a unique car logo and to improve detection of car logos in Indonesia. The logo that was later discovered was used to find the car brand in no time. In this study we use the Single Shot Multibox Detector method which is known to detect objects running on the Jupyter Notebook Application. The data used for this research is of a public nature obtained from the Kaggle website dataset source which contains a number of varying images. There are 7 classes of car brands, namely Volkswagen, Hyundai, Lexus, Mercedes, Peugeot, Renault, and Tesla. Data testing in this study obtained 1,240 images for training data and 270 images in the test data category that had been tested and resulted in an evaluation value with the best accuracy value of 98.89% and a loss value of 0.025%.

show abstract

Section: Model Terlatihunclassified

Optimasi Mendeteksi Klasifikasi Citra Digital Logo Mobil Indonesia Dengan Metode Single Shot Multibox Detector

Mulyana¹,

Zikri²

2022

Explore. jurnal. sistem. inf. dan. telematika

View full text Add to dashboard Cite

show abstract

“…here the image is divided into matrix cells, the count of which is predetermined and each of these cells are a proposal. Improved model of Yolo, V2 and V3 [41] [42] [43] shows improvement in accuracy and speed of detection and is used extensively for object detection in videos [43][44][45][46] though there are underlying issues of localization in this algorithm. YOLO is faster compared to the region based models but accuracy is not on par with the latter especially in distinguishing objects of smaller size and images with different aspect ratios.…”

Section: Figure 17 Classification Of the Techniques For Vehicle And On-road Object Detectionmentioning

confidence: 99%

“…The Model Predictive Controllers tend to optimize the algorithms and hence reduce the cost overhead. The nonlinear vehicle based MPC [72], extended kinematic based model [45] are the most prevalent models. Robust Controllers [73][74][75][76] though highly complex and robust in their design have the ability to negate the unpredictable changes in dynamically changing conditions.…”

Section: Figure 19 Major Vehicle Models Used For Path Predictionmentioning

confidence: 99%

Adaptive Headlamps in Automobile: A Review on the Models, Detection Techniques, and Mathematical Models

Toney

Bhargava

2021

IEEE Access

View full text Add to dashboard Cite

Driving at night with traditional headlamps poses great threats with a large number of accidents occurring during the night because of temporary blindness caused by the headlights of the oncoming traffic. The headlights when in high beam cause temporary visual impairment of human eyes called the Troxler effect. While it reduces the time to react, it also leads to decreased visibility which contributes to the majority of mishaps that occur at night. Customarily the headlight adjustments are controlled manually where poor driving skills or error in judgment can have catastrophic effects. Accidents also occur due to poor lighting conditions as the current regular headlamp configurations don't illuminate the roads precisely, especially during curves and on unpredictable terrains. Hence there is a need for adaptive headlamps in automobiles that can prevent Troxler's effect on the drivers of the opposite vehicles while not compromising on the illumination of the road for the driver on-board. This paper explores various methodologies used in implementing adaptive headlamps and explore the scope for future work in this area of research. Also, the paper reviews vehicle detection algorithms as well as various vehicle mathematical models for headlamp control based on steering angles INDEX TERMS Adaptive Front light Systems (AFS)

show abstract

“…Biswas et al [5] implemented SSD to estimate traffic density. Gupta et al [6] analyzed the two algorithms YOLOv3 and SSD to count the number of people at any junction. Their results have shown that SSD is better than YOLOV3 v3.…”

Section: Introductionmentioning

confidence: 99%

Detection of Lower Body for AGV Based on SSD Algorithm with ResNet

Gao

Luo

et al. 2022

Sensors

View full text Add to dashboard Cite

Detection of human lower body provides an implementation idea for the automatic tracking and accurate relocation of automatic vehicles. Based on traditional SSD and ResNet, this paper proposes an improved detection algorithm R-SSD for human lower body detection, which utilizes ResNet50 instead of VGG16 to improve the feature extraction level of the model. According to the application of acquisition equipment, the model input resolution is increased to 448 × 448 and the model detection range is expanded. Six feature maps of the updated resolution network are selected for detection and the lower body image dataset is clustered into five categories for aspect ratio, which are evenly distributed to each feature detection map. The experimental results show that the model R-SSD detection accuracy after training reaches 85.1% mAP. Compared with the original SSD, the detection accuracy is improved by 7% mAP. The detection confidence in practical application reaches more than 99%, which lays the foundation for subsequent tracking and relocation for automatic vehicles.

show abstract

WITHDRAWN: People detection and counting using YOLOv3 and SSD models

Cited by 14 publications

References 18 publications

Optimasi Mendeteksi Klasifikasi Citra Digital Logo Mobil Indonesia Dengan Metode Single Shot Multibox Detector

Optimasi Mendeteksi Klasifikasi Citra Digital Logo Mobil Indonesia Dengan Metode Single Shot Multibox Detector

Adaptive Headlamps in Automobile: A Review on the Models, Detection Techniques, and Mathematical Models

Detection of Lower Body for AGV Based on SSD Algorithm with ResNet

Contact Info

Product

Resources

About