VAVeysel Aslantas

Veysel Aslantas

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Veysel Aslantas has a total of 17 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    SPDNet: Spliced Person Detection Network Using Two-Channel Four-Stream CNN Architecture

    by Mehmet Elmaci, Ahmet Nusret Toprak, Veysel Aslantas

    Image splicing, particularly inserting individuals into scenes, presents significant challenges for detecting image forgery. This paper introduces SPDNet, a two-stage deep learning framework for detecting spliced persons. The first stage segments all person instances using a PointRend-based model, while the second stage analyzes each instance using a two-channel, four-stream architecture that fuses RGB and SRM noise features via EfficientNet-B4 backbones and Squeeze-and-Excitation modules. We evaluated SPDNet on the DSO-1 and OIS datasets as well as our newly introduced SPD Evaluation dataset, comparing it against eight state-of-the-art image splicing detection models. Experimental results show that SPDNet outperforms existing methods in both accuracy and localization of spliced individuals, demonstrating strong potential for person-level image forensics.


  • Open Access

    A Novel Self-Embedded Dual Watermarking Method for Image Tamper Detection and Recovery

    by Huseyin Ozkaya, Veysel Aslantas

    Digital images have been used frequently in our daily lives. Manipulations on images can easily be made through various image-editing software. Detecting and recovering these manipulations are an important issue. In this paper, a novel self-embedded dual watermarking method based on discrete wavelet transform for tamper detection and recovery is proposed. This method is a block-based approach. It divides the input image into non-overlapping blocks. The watermark used for each block is constructed using recovery and authentication information that are obtained from the considered wavelet transformed block. The recovery information is generated from the seven MSBs of the LL1 coefficient. The first five bits of the LL1 and block number are used to generate the authentication bits. The authentication bits of each block are embedded into the block itself, while the recovery bits are embedded in two different blocks determined from the lookup table. The advantage of embedding the same information of a block into two different blocks is that when one of the blocks is damaged, the information can be recovered from the other one. The developed technique is evaluated under various attacks such as cropping and collage and compared with several state-of-the-art methods. Experimental results show that the proposed approach can easily detect the tempered areas of the watermarked image against various attacks and effectively recover those areas using the self-embedded dual watermark. The results obtained certify the considerable accuracy and feasibility of the proposed technique.


  • Open Access

    An Improved Marine Predators Algorithm for Shipment Status Time Estimation and Regression Problems

    by Resul Özdemir, Murat Tasyurek, Veysel Aslantas

    In the realm of machine learning, the optimization of hyperparameters is a pivotal step in creating intelligent models of high efficacy. Configuration of hyperparameters in a smart way contributes to enhanced model generalization and resilience, mitigating the risks associated with overfitting or underfitting. Optimization algorithms play a decisive role in determining the ideal hyperparameters for machine learning algorithms. This study introduces a novel hyperparameter optimization approach based on the Marine Predators Algorithm designed for regression problems. The performance of the proposed algorithm is benchmarked against a spectrum of popular and classical optimization algorithms from the literature. Furthermore, hyperparameter optimization is examined for three widely employed machine learning algorithms which are Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM) and Support Vector Regression (SVR). Regression models are constructed using a actual world dataset for prediction of shipment status time, and the proposed algorithm is further evaluated with public three datasets suitable for regression analysis. A new hyperparameter decoding approach is proposed to turn hyperparameter optimization into a numerical optimization problem regardless of parameter type. Consequently, the experimental results indicate that the proposed optimization algorithm exhibited superior performance than the compared optimizers in hyperparameter optimization for both machine learning algorithms across all regression datasets.

  • Open Access

    A Triple Self-Embedding Fragile Watermarking Scheme for Image Tamper Detection and Recovery

    by Huseyin Ozkaya, Veysel Aslantas

    This paper presents a new triple self-embedding fragile watermarking technique for image tamper detection and recovery. This approach divides the host image into non-overlapping regions according to the block map and further divides each region into 5×5 non-overlapping blocks. For each block, block directions are obtained from the adjacent pixel differences. The recovery information for each block is constructed by computing the block directions, differences of the directions and the mean value of each block. The authentication information for each block is generated using an Singular Value Decomposition (SVD)-based approach. The recovery information is embedded in three different blocks determined using a block map while authentication information is embedded in the block itself using the three least significant bits. The developed approach provides a triple recovery opportunity for each block due to the triple embedding. Consequently, the watermark constructing and embedding processes of this approach provide more successful image tamper recovery than the methods based on average values of blocks. The proposed approach has been evaluated against various attacks and compared with state-of-the-art methods. Experimental results shows that the proposed approach can resist attacks up to 75%.


  • Open Access

    A Fast and Accurate Method for Classifying Tomato Plant Health Status Using Machine Learning and Image Processing

    by Hasan Ulutas, Veysel Aslantas

    Agriculture is crucial to economic growth and development, and maintaining high-quality, disease-free plants is crucial to its success. Early detection of plant diseases, which can be caused by environmental factors, fungi, bacteria, and viruses, is essential to implement appropriate treatments. Tomatoes, which are one of the most vital food crops, are susceptible to diseases that can result in significant economic losses in agriculture. This study introduces a method to evaluate the health of tomato leaf using image processing techniques and machine learning algorithms. A dataset of 1,778 images of healthy and infected tomato leaves was collected from tomato planting areas in the Turkish provinces of Samsun and Mersin. Sixteen advanced machine learning algorithms were used for classification, and the optimal hyperparameters for each algorithm were determined using a grid search approach. The classifiers were executed on Jetson Nano and TX2 embedded systems. The experimental results indicate that the Random Forest classifier outperformed other algorithms, achieving approximately 99 % accuracy in detecting and classifying the health status of tomato leaves. The proposed system enables faster and more accurate detection, allowing farmers to classify plants as infected or healthy, ultimately improving decision-making on treatment and pest management strategies.


  • Open Access

    Design of Efficient Methods for the Detection of Tomato Leaf Disease Utilizing Proposed Ensemble CNN Model

    by Hasan Ulutaş, Veysel Aslantaş

    Early diagnosis of plant diseases is of vital importance since they cause social, ecological, and economic losses. Therefore, it is highly complex and causes excessive workload and time loss. Within the scope of this article, nine tomato plant leaf diseases as well as healthy ones were classified using deep learning with new ensemble architectures. A total of 18.160 images were used for this process. In this study, in addition to the proposed two new convolutional neural networks (CNN) models, four other well-known CNN models (MobileNetV3Small, EfficientNetV2L, InceptionV3 and MobileNetV2) are used. A fine-tuning method is applied to the newly proposed CNNs models and then hyperparameter optimization is performed with the particle swarm optimization algorithm (PSO). Then, the weights of these architectures are optimized by the grid search method and triple and quintuple ensemble models are created and the datasets are classified with the help of the five-fold cross-validation. The experimental results demonstrate that the proposed ensemble models stand out with their fast training and testing time and superior classification performances with an accuracy of 99.60%. This research will help experts enable the early detection of plant diseases in a simple and quick manner and prevent the formation of new infections.


  • Open Access

    Evrişimli sinir ağı kullanarak çoklu-pozlamalı görüntü birleştirme

    by Harun Akbulut, Veysel Aslantaş

    Aynı sahneye ait iki ya da daha fazla düşük dinamik alana (LDR) sahip görüntülerden yüksek dinamik alana (HDR) sahip tek bir görüntü elde etme yöntemlerine çoklu-pozlamalı görüntü birleştirme (MEF) denir. Bu çalışmada MEF için derin öğrenme (DL) modellerinden evrişimli sinir ağı (CNN) kullanan yeni bir yöntem önerilmiştir. Önerilen yöntemde ilk adımda CNN modeli kullanılarak kaynak görüntülerden birleştirme haritası (fmap) elde edilmiştir. Birleştirilmiş görüntülerde testere-dişi etkisini ortadan kaldırmak için fmap üzerinde ağırlıklandırma işlemi gerçekleştirilmiştir. Daha sonra ağırlıklandırılmış fmap kullanılarak her tarafı iyi pozlanmış birleştirilmiş görüntüler oluşturulmuştur. Önerilen yöntem literatürde yaygın olarak kullanılan MEF veri setlerine uygulanmış ve elde edilen birleştirilmiş görüntüler kalite metrikleri kullanılarak değerlendirilmiştir. Önerilen yöntem ve diğer iyi bilinen görüntü birleştirme yöntemleri, görsel ve niceliksel değerlendirme açısından karşılaştırılmıştır. Elde edilen sonuçlar, geliştirilen tekniğin uygulanabilirliğini göstermektedir.