Multimodel Deep Learning for Person Detection in Aerial Images

Vasić, Mirela Kundid; Papić, Vladan

doi:10.3390/electronics9091459

Cited by 27 publications

(7 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some findings and results obtained during the development of such an expert system are already presented, but they relate to the transmission of images from UAVs to the ground station [16] and novel algorithms for person detection in aerial images [24,25].…”

Section: Introductionmentioning

confidence: 99%

Person Mobility Algorithm and Geographic Information System for Search and Rescue Missions Planning

Papić,

Gudelj,

Milan

et al. 2024

Remote Sensing

Self Cite

View full text Add to dashboard Cite

In search and rescue (SAR) operations, up-to-date information on the terrain is critical because every additional hour required to search for a person reduces the likelihood of success. Therefore, it is necessary to provide quick access and the best possible input data for planners and search teams and to develop tools that can help them plan and monitor actions in real-time. This paper describes a novel system based on the use of GIS for planning actions and visualizing the situation on the ground. Special focus is devoted to the algorithm for assessing the mobility of the missing person. Using this algorithm, the area of the proposed search area is modeled based on obtained information about the type of terrain on which the searches are planned. The obtained results are presented as a new Geographic Information System layer and have proven to be a quality that helps in defining the search space. Further research is discussed, especially regarding the assessment of the passability of certain types of terrain.

show abstract

Section: Introductionmentioning

confidence: 99%

Person Mobility Algorithm and Geographic Information System for Search and Rescue Missions Planning

Papić,

Gudelj,

Milan

et al. 2024

Remote Sensing

Self Cite

View full text Add to dashboard Cite

show abstract

“…Deep learning has been successfully applied to a variety of fields, including computer vision and machine vision (Gholape et al, 2021;Solaiman et al, 2022;Kundid Vasić and Papić, 2020). Face recognition techniques are particularly useful for searching for missing people, as they save time and effort.…”

Section: Introductionmentioning

confidence: 99%

Suhail: A Deep Learning-Based System for Identifying Missing People

Aljohani¹,

Alshehri²,

Alghamdi³

et al. 2023

CIS

View full text Add to dashboard Cite

Many people become missing in Saudi Arabia every day, including children, young people, and the mentally ill as well as the elderly with Alzheimer’s. There are many missing people cases that are still unsolved. In Saudi Arabia, people use social media platforms such as Twitter to report missing people cases. The application of deep learning has been successful in a wide range of fields including computer vision and machine vision. In particular, face recognition techniques are effective in saving time and effort, especially when searching for missing individuals. Hence, the goal of this research is to solve this issue by developing a deep learning-based system for identifying missing individuals. This paper introduces a new system called Suhail. The system has been implemented and developed using Android Studio and open-source libraries such as TensorFlow. First, users or governments can report missing persons by uploading photos. Updates and information will then be shared with the rest of the system’s users (volunteers). Once a volunteer discovers a suspect, they scan their face using camera. Then, our application uses face recognition techniques to compare the suspect's photo with photos from the repository. Finally, once a comparison is found, our application contacts the suspect’s family, informs them of his location and then notifies the police that a missing person has been located. By using our application and face recognition systems, we help families and police locate and reach a missing person which saves time and effort. In this study, 759 participants were enrolled to evaluate the performance of the Suhail system. Engagement, aesthetics, and functionality are used to evaluate the user experience. The results of the experiment show that users enjoy the new features of the application and that the system is simple to use. Moreover, the system would help governments and individuals identify missing people faster.

show abstract

“…Researchers have attempted to create datasets of persons in aerial images and have obtained some promising results by utilizing existing creative object detection algorithms based on natural images. However, the results have been weak and lack robustness [5][6][7][8]. Compared to remote sensing objects [9][10][11][12][13][14][15], such as ships, vehicles, and airplanes, persons in aerial images are frequently costly to identify in SaR scenes, difficult to label manually, have fewer available datasets, and have multi-view shooting features that vary greatly.…”

Section: Introductionmentioning

confidence: 99%

Semi-Supervised Person Detection in Aerial Images with Instance Segmentation and Maximum Mean Discrepancy Distance

et al. 2023

View full text Add to dashboard Cite

Detecting sparse, small, lost persons with only a few pixels in high-resolution aerial images was, is, and remains an important and difficult mission, in which a vital role is played by accurate monitoring and intelligent co-rescuing for the search and rescue (SaR) system. However, many problems have not been effectively solved in existing remote-vision-based SaR systems, such as the shortage of person samples in SaR scenarios and the low tolerance of small objects for bounding boxes. To address these issues, a copy-paste mechanism (ISCP) with semi-supervised object detection (SSOD) via instance segmentation and maximum mean discrepancy distance is proposed (MMD), which can provide highly robust, multi-task, and efficient aerial-based person detection for the prototype SaR system. Specifically, numerous pseudo-labels are obtained by accurately segmenting the instances of synthetic ISCP samples to obtain their boundaries. The SSOD trainer then uses soft weights to balance the prediction entropy of the loss function between the ground truth and unreliable labels. Moreover, a novel evaluation metric MMD for anchor-based detectors is proposed to elegantly compute the IoU of the bounding boxes. Extensive experiments and ablation studies on Heridal and optimized public datasets demonstrate that our approach is effective and achieves state-of-the-art person detection performance in aerial images.

show abstract

Multimodel Deep Learning for Person Detection in Aerial Images

Cited by 27 publications

References 39 publications

Person Mobility Algorithm and Geographic Information System for Search and Rescue Missions Planning

Person Mobility Algorithm and Geographic Information System for Search and Rescue Missions Planning

Suhail: A Deep Learning-Based System for Identifying Missing People

Semi-Supervised Person Detection in Aerial Images with Instance Segmentation and Maximum Mean Discrepancy Distance

Contact Info

Product

Resources

About