A5087637805- Industrial Vision Systems and Defect Detection
- Advanced Steganography and Watermarking Techniques
- Digital Media Forensic Detection
- Music and Audio Processing
- Video Surveillance and Tracking Methods
- Microwave Imaging and Scattering Analysis
- Indoor and Outdoor Localization Technologies
- Advanced Statistical Process Monitoring
- Welding Techniques and Residual Stresses
- Chaos-based Image/Signal Encryption
- Manufacturing Process and Optimization
- Advanced Neural Network Applications
- Advanced Vision and Imaging
- Energy Load and Power Forecasting
- Sparse and Compressive Sensing Techniques
- Power Quality and Harmonics
- Power System Reliability and Maintenance
- Visual Attention and Saliency Detection
University of Montenegro
2020-2024
Manufacturing companies focus on improving productivity, reducing costs, and aligning performance metrics with strategic objectives. In industries like paper manufacturing, minimizing equipment downtime is essential for maintaining high throughput. Leveraging the extensive data generated by these facilities offers opportunities gaining competitive advantages through data-driven insights, revealing trends, patterns, predicting future indicators unplanned length, which in optimizing...
Visual inspection plays a pivotal role in numerous industrial production processes, and the pursuit of automation has surged with rise deep learning convolutional neural networks (CNNs). Therein, deployment visual CNNs on resource-constrained edge devices stands as critical problem these are most affordable well-suited for many applications, e.g., chains. Nonetheless, it faces challenges meeting computational demands CNN models. Consequently, optimizing models efficient operation such...
Desynchronization attacks proved to be the greatest challenge audio watermarking systems as they introduce misalignment between signal carrier and watermark. This paper proposes a DNN-based speech system with two adversarial networks jointly trained on set of desynchronization embed randomly generated The detector neural network is expanded spatial pyramid pooling layers able handle signals affected by these attacks. A detailed training procedure aforementioned DNN gradual attack...
This paper implements the task of semi-supervised Video Object Segmentation (VOS), i.e., separation an object from background in a video, given mask first frame. To accomplish this task, modern Machine learning techniques have been used, such as, Convolution Neural Networks (CNNs) and Convolutional Recurrent (CRNNs). The motion objects between consecutive frames sequence, caused by relative movement camera is very important information bears name Optical Flow (OF). OF was used to improve...
Watermarking is a process in which both physical and digital media are marked using watermarks order to protect ownership of the watermarked media. Digital watermarking technique where watermark gets embedded into carrier signal while preserving quality original Embedding can happen various domains could be hidden plain, but information carried by should not deteriorate. This paper deals with hiding speech audio signals deep neural networks. We present an encoder-decoder architecture that...
This paper deals with the Compressive Sensing implementation in Face Recognition problem. is new approach signal processing a single goal to recover from small set of available samples. finds its usage many real applications as it lowers memory demand and acquisition time, therefore allows dealing huge data fastest manner. In this paper, undersampled recovered using algorithm based on Total Variation minimization. The theory verified an experimental results different percentage
This paper presents the results of applying optimization techniques, most notably neural architecture search (NAS) and hyperparameter (HPO) strategies, to a known state-of-the-art deep learning model for surface defect detection in industry. It will be shown that it is possible achieve significant reduction latency its number parameters, while incurring only negligible drop accuracy. The main motivation this was deployment models on edge devices with very limited computational capabilities,...