17 publications from 2015 to 2026: 2 journal articles, 10 conference papers, 1 book chapter and 4 preprints, on accessible navigation, wildfire and environmental AI, biomedical image analysis, and efficient learning.
Under review · arXiv preprint arXiv:2609.17763 (2026)PreprintUnder review
Modular deep learning augmentations for next-day wildfire spread prediction: wind- and slope-conditioned attention, retrieval-augmented correction, and dual-stream gating.
IEEE International Symposium on Workload Characterization (IISWC) 2026Conference paper
An end-to-end characterization of an on-device, depth-aware multimodal workload for assistive indoor navigation for blind and low-vision users.
CVPR 2026 Workshop (GRAIL-V) · IEEE/CVF Conference on Computer Vision and Pattern Recognition WorkshopsConference paper
An agentic framework that converts a single floor plan image into a structured, retrievable knowledge base for safe, accessible indoor navigation.
IEEE International Geoscience and Remote Sensing Symposium (IGARSS) 2026Conference paper
A vision-language model for early wildfire detection and risk assessment from satellite imagery.
2025 IEEE International Conference on Big Data (BigData)Conference paper
Floorplan2Guide transforms floorplans into spatial knowledge graphs and language-guided indoor navigation for blind and low-vision users.
Expert Systems with Applications, vol. 213, 119040, 2023Journal article
A clustering-based hint selection method for knowledge distillation: teacher layers are clustered and the cluster centers become the hint points.
In Diagnostic Biomedical Signal and Image Processing Applications with Deep Learning Methods, pp. 137-154, Academic Press, 2023Book chapter
An end-to-end workflow for analyzing phase-contrast microscopy data from cell motility and wound healing assays.
Turkish Journal of Electrical Engineering and Computer Sciences, vol. 29, no. 8, pp. 2855-2868, 2021Journal article
A modified U-Net with a pretrained ResNet-18 encoder, residual decoder blocks, and a redesigned skip-connection for cell segmentation in label-free microscopy images.
arXiv preprint arXiv:2105.00695 (2021)Preprint
Residual connections for deep graph autoencoders: ResVGAE, a deep variational graph autoencoder with multiple residual modules, for link prediction.
2020 28th Signal Processing and Communications Applications Conference (SIU), pp. 1-4, IEEE, 2020Conference paper
A deep learning segmentation pipeline for label-free phase-contrast microscopy images, improving U-Net and FPN with an ImageNet-pretrained ResNet-18 encoder.
2020 28th Signal Processing and Communications Applications Conference (SIU), pp. 1-4, IEEE, 2020Conference paper
Residual Variational Graph Autoencoder (RVGAE): residual connections boost variational graph autoencoders for link prediction.
arXiv preprint arXiv:1904.09209 (2019)Preprint
A new class of affine scaling matrices for interior point Newton-type methods, defined as a convex combination of well-known scaling matrices.
2019 Medical Technologies Congress (TIPTEKNO), pp. 1-4, IEEE, 2019Conference paper
A multiresolution U-Net with multi-combination augmentation for cell segmentation in 2D phase-contrast microscopy images.
2019 Medical Technologies Congress (TIPTEKNO), pp. 1-4, IEEE, 2019Conference paper
Traditional image processing and deep learning approaches for automated cell segmentation from phase-contrast microscopy time series, evaluated against manually annotated datasets.
Preprints (2018), doi:10.20944/preprints201812.0232.v1Preprint
Fine-tuned pretrained ImageNet models for dog breed identification, reaching 89.66% test accuracy on the Stanford Dogs dataset.
2015 2nd International Conference on Knowledge-Based Engineering and Innovation (KBEI), pp. 504-509, IEEE, 2015Conference paper
A modified ant colony optimization method for digital image edge detection, faster and more efficient than former ant colony based methods.
International Conference on New Research Findings in Electrical Engineering and Computer Science, Tehran, 2015 · arXiv:1710.01260Conference paper
A support vector machine with a three-dimensional Gaussian radial basis function kernel for edge detection, outperforming Canny and Sobel detectors.
No publications match those filters.
Citation counts are on Google Scholar.