Comparative Analysis of Deep Learning Image Detection Algorithms

Srivast, Shrey; Divekar, Amit Vishvas; Anilkumar, Chandu; Naik, Ishika; Kulkarni, Ved; Pattabiraman, V.

doi:10.21203/rs.3.rs-132774/v1

Cited by 13 publications

(12 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…At the current moment, there are five versions of YOLO networks. As mentioned in the article by Srivastava et al [32], YOLOv3 allows for rapid calculation while maintaining very high accuracy. The transition from DarkNet to the PyTorch with the fifth version allowed for the rapid development of YOLOv5 [33], which started with lower accuracy.…”

Section: Related Methodsmentioning

confidence: 99%

Real-Time Object Detection and Classification by UAV Equipped With SAR

Gromada

Siemiątkowska

Stecz

et al. 2022

Sensors

View full text Add to dashboard Cite

The article presents real-time object detection and classification methods by unmanned aerial vehicles (UAVs) equipped with a synthetic aperture radar (SAR). Two algorithms have been extensively tested: classic image analysis and convolutional neural networks (YOLOv5). The research resulted in a new method that combines YOLOv5 with post-processing using classic image analysis. It is shown that the new system improves both the classification accuracy and the location of the identified object. The algorithms were implemented and tested on a mobile platform installed on a military-class UAV as the primary unit for online image analysis. The usage of objective low-computational complexity detection algorithms on SAR scans can reduce the size of the scans sent to the ground control station.

show abstract

Section: Related Methodsmentioning

confidence: 99%

Real-Time Object Detection and Classification by UAV Equipped With SAR

Gromada

Siemiątkowska

Stecz

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…The original resolution of the smartphone image is 4032 x 3024 pixels which are resized to 416 x 416 pixels. It aims to make the training process lighter and faster [9]. The resizing process is done using the Pillow library in Python 3.7.…”

Section: Image Collectionmentioning

confidence: 99%

“…In this study, the YOLO algorithm was chosen because it has the best processing speed, which is crucial when processing real-time images and videos [9]. Specifically, this research uses the 4th version (YOLOv4), the state-of-art of the YOLO algorithm.…”

Section: Introductionmentioning

confidence: 99%

Real-time detection of fruit ripeness using the YOLOv4 algorithm

Widyawati¹,

Febriani²

2021

tjst

View full text Add to dashboard Cite

Indonesia has an abundance and variety of fruit commodities. Referring to BPS, in 2019 the fruit production reached 22 million tons or increase 5% compared to 2018. However, with such a large production volume, most of the fruit inspection process in Indonesia still performed with human intervention. This process is very labor intensive, time consuming, and prone to inconsistencies and inaccuracies. The automation process using computer vision technology is expected to eliminate manual process therefore reducing costs, increasing efficiency and accuracy. In this study, YOLOv4 algorithm was applied to detect banana ripeness automatically. The training process is carried out using 369 banana images which are divided into two classes and the testing process is carried out on videos that are captured in real-time. Based on the research results, the best average accuracy rate is 87.6% and the video processing speed is 5 FPS (frames per second) using a single-GPU architecture.

show abstract

“…In this research [16], different pre-trained models using two datasets MS COCO [17] and PASCAL VOC [18] have been reviewed for object detection such as R-CNN, R-FCN, SSD, and YOLO, with different feature extractors such as VGG16, ResNet, Inception, MobileNet. Srivastava et al [19] were presented a comparative analysis of 3 major image processing algorithms: SSD, Faster R-CNN, and YOLO. In this analysis, they chose the COCO dataset to evaluate the performance and accuracy of the three algorithms and analyzed their strengths and weaknesses.…”

Section: Intputmentioning

confidence: 99%

Object Detection Approaches in Images: A Weighted Scoring Model based Comparative Study

Ouchra¹,

Belangour²

2021

IJACSA

View full text Add to dashboard Cite

Computer vision is a branch of artificial intelligence that trains computers to acquire high-level understanding of images and videos. Some of the most wellknown areas in Computer Vision are object detection, object tracking and motion estimation among others. Our focus in this paper concerns object detection subarea of computer vision which aims at recognizing instances of predefined sets of objects classes using bounding boxes or object segmentation. Object detection relies on various algorithms belonging to various families that differs in term of speed and quality of results. Hence, we propose in this paper to provide a comparative study of these algorithms based on a set of criteria. In this comparative study we will start by presenting each of these algorithms, selecting a set of criteria for comparison and applying a comparative methodology to get results. The methodology we chose to this purpose is called WSM (Weighted Scoring Model) which fits exactly our needs. Indeed, WSM method allows us to assign a weight to each of our criterion to calculate a final score of each of our compared algorithms. The obtained results reveal the weaknesses and the strengths of each one of them and opened breaches for their future enhancement.

show abstract

Comparative Analysis of Deep Learning Image Detection Algorithms

Cited by 13 publications

References 13 publications

Real-Time Object Detection and Classification by UAV Equipped With SAR

Real-Time Object Detection and Classification by UAV Equipped With SAR

Real-time detection of fruit ripeness using the YOLOv4 algorithm

Object Detection Approaches in Images: A Weighted Scoring Model based Comparative Study

Contact Info

Product

Resources

About