A5013504831- Advanced Image and Video Retrieval Techniques
- Sports Performance and Training
- Robotics and Sensor-Based Localization
- Video Surveillance and Tracking Methods
- Sports Analytics and Performance
- Advanced Vision and Imaging
- Physical Education and Training Studies
- Video Analysis and Summarization
- Infrared Target Detection Methodologies
- Hand Gesture Recognition Systems
- Advanced Neural Network Applications
- Remote-Sensing Image Classification
- Teaching and Learning Programming
- Robotics and Automated Systems
- Human Pose and Action Recognition
- Anomaly Detection Techniques and Applications
- Robot Manipulation and Learning
- Gait Recognition and Analysis
- Indoor and Outdoor Localization Technologies
- Robotic Path Planning Algorithms
- Regional Development and Management Studies
- Genetics and Physical Performance
- Sparse and Compressive Sensing Techniques
- Medical Image Segmentation Techniques
- Computer Graphics and Visualization Techniques
University of Split
2015-2024
Tehničko Veleučilište u Zagrebu
2010
Compressive sensing has emerged as an area that opens new perspectives in signal acquisition and processing. It appears alternative to the traditional sampling theory, endeavoring reduce required number of samples for successful reconstruction. In practice, compressive aims provide saving resources, transmission, storage capacities facilitate processing circumstances when certain data are unavailable. To end, relies on mathematical algorithms solving problem reconstruction from a greatly...
In this paper, we propose a novel method for person detection in aerial images of nonurban terrain gathered by an Unmanned Aerial Vehicle (UAV), which plays important role Search And Rescue (SAR) missions. The UAV SAR operations contributes significantly due to the ability survey larger geographical area from viewpoint. Because high altitude recording, object interest (person) covers small part image (around 0.1%), makes task quite challenging. To address problem, multimodel deep learning...
In this letter, a system combining compressive sensing (CS)-based image reconstruction and object detection algorithm is introduced. The use of CS promising approach for search-and-rescue applications, since it highly reduces the amount data that needs to be transmitted. However, high-quality such images challenging task due complexity structures number tiny details, possibly being objects interest. Hence, performance evaluated in terms missing quality. Object performed by applying two-stage...
Recent results in person detection using deep learning methods applied to aerial images gathered by Unmanned Aerial Vehicles (UAVs) have demonstrated the applicability of this approach scenarios such as Search and Rescue (SAR) operations. In paper, continuation our previous research is presented. The main goal further improve results, especially terms reducing number false positive detections consequently increasing precision value. We present a new that, input multimodel neural network...
The aim of saliency object detection algorithms is to find objects in image or video which draw attention humans at the first sight. This very popular topic robotics and computer vision research useful various areas applications like segmentation, adaptive compression, recognition, visual surveillance so on. In this paper, we will explore possibilities using these on problem for Search Rescue operations (SAR) UAV images.
Search and rescue operations usually require significant resources, personnel, equipment, time. In order to optimize the resources expenses increase efficiency of operations, use unmanned aerial vehicles (UAVs) photography is considered for fast reconnaissance large unreachable terrains. The images are then transmitted control center automatic processing pattern recognition. Furthermore, due limited transmission capacities battery consumption recording high resolution images, in this paper...
Marine litter poses a significant global threat to marine ecosystems, primarily driven by poor waste management, inadequate infrastructure, and irresponsible human activities. This research investigates the application of image preprocessing techniques deep learning algorithms for detection seafloor objects, specifically debris, using unmanned aerial vehicles (UAVs). The primary objective is develop non-invasive methods detecting mitigate environmental impacts support health ecosystems. Data...
Search and rescue (SAR) missions comprise search for, provision of aid to people who are in distress or imminent danger. Providing the best possible input for planners teams, up-to-date information about terrain is essential importance because every additional hour needed a person decreases probability success. Therefore, availability aerial images updated maps as basis planning monitoring SAR real-time very important rescuers. In this paper, we present system transmission high-resolution...
In search and rescue (SAR) operations, up-to-date information on the terrain is critical because every additional hour required to for a person reduces likelihood of success. Therefore, it necessary provide quick access best possible input data planners teams develop tools that can help them plan monitor actions in real-time. This paper describes novel system based use GIS planning visualizing situation ground. Special focus devoted algorithm assessing mobility missing person. Using this...
This paper presents a novel two-stage image segmentation approach for detection of artificial materials and objects in non-urban terrain. It is based on the assumption that an object natural background have different color/saturation. High resolution unknown terrain taken with digital camera from distance about 100 m divided into smaller sub-images further processing. Each sub-image clustered using mean shift algorithm, information obtained cluster centers transferred to next stage. Second...
The discrimination of ECG signals is crucial importance in clinical diagnoses cardiac diseases. Manual analysis very complex and time consuming task due to their composite nature. This paper proposes a novel scheme for reliable automatic classification into normal three different abnormal (arrhythmia affected) categories. feature extraction based on an amalgamation discrete cosine transform random projection dimensionality reduction. Furthermore, the performed using forest algorithm....
In this paper, a novel approach for an automatic object detection and localisation on aerial images is proposed. Proposed model does not use ground control points (GCPs) consists of three major phases. the first phase, optimal flight route planned in order to capture area interest are acquired using unmanned vehicle (UAV), followed by creating mosaic collected obtained larger field-of-view panoramic image create georeferenced map. The then also used detect objects based convolutional neural networks.
The need to improve communication between humans and computers has been motivation for defining new models, accordingly, ways of interacting with machines. In many applications today, user interaction is moving away from traditional keyboards mouses becoming much more physical, pervasive intuitive. This paper examines hand gestures as an alternative or supplementary input modality mobile devices. A gesture recognition system based on the use acceleration sensor, that nowadays being featured...
In this paper, an automatic method for detecting and counting olive trees in RGB images acquired by unmanned aerial vehicle (UAV) is developed. Our approach based on implementation of RetinaNet model DeepForest Phyton package. For improvement pretrained via transfer learning, five groves were mapped using UAV, manually labeled, new image dataset was created. Several models built, each being trained evaluated different set from selected groves. Experimental results obtained a UAV over test...
Automated solutions for medical diagnosis based on computer vision form an emerging field of science aiming to enhance and early disease detection. The detection quantification facial asymmetries enable palsy evaluation. In this work, a detailed review the takes place, covering all methods ranging from traditional manual mathematical modeling automated vision-based methods. Moreover, is defined in terms asymmetry indices calculation different image modalities. aim introduce readers concept...