feat(content): add alumni entries and update publications

- Add alumni profiles for Dağlar Berk Erdem and Oğulcan Özdemir
- Update Berk Gökberk and Lale Akarun BibTeX up to early 2023
- Regenerate/add paper content entries from updated BibTeX

Author: kutayeroglu

data/bib/berk-gokberk.bib

@@ -1,3 +1,71 @@
+@inproceedings{girgin2023novel,
+  title={A novel occlusion index},
+  author={Girgin, Emre and G{\"o}kberk, Berk and Akarun, Lale},
+  booktitle={2023 31st Signal Processing and Communications Applications Conference (SIU)},
+  pages={1--4},
+  year={2023},
+  organization={IEEE},
+  abstract={Recovery of 3D human pose and shape under realistic conditions is a challenging task. Despite the recent advances in this field, methods suffer from performance degradation due to naturally occurring occlusions. Benchmark datasets employed to compare the performance of methods under occlusion contain different amounts of challenges. Therefore, assessing the quantitative performance of the benchmarks turns into an equally important task. In this study, we propose a novel metric called the Occlusion Index (OI) to evaluate the severity of occlusion for benchmarks. OI enables the evaluation of the amount of occlusion and partitioning of the benchmark images into subsets according to the severity of the occlusion. We provide occlusion benchmarks obtained from widely-utilized datasets, OCHuman and AGORA, and show that they enable better evaluation of the occlusion robust techniques.},
+  url={https://ieeexplore.ieee.org/abstract/document/10223983}
+}
+
+@inproceedings{yildiz2023survey,
+  title={A survey of deepfake detection methods},
+  author={Yildiz, Burak {\.I}kan and G{\"o}kberk, Berk},
+  booktitle={2023 31st Signal Processing and Communications Applications Conference (SIU)},
+  pages={1--4},
+  year={2023},
+  organization={IEEE},
+  abstract={In this work, we examine recent methods employed for detecting deepfake images and videos, which have become increasingly prevalent in recent times. Deepfake datasets are categorized into generations based on the number of data, quality, and diversity of the data they contain, and a new realistic dataset has been generated to accurately reflect real-world conditions. In this paper, we evaluate the methodologies used for deepfake detection using computer vision and deep learning on currently available datasets and present the results in a comprehensive manner.},
+  url={https://ieeexplore.ieee.org/abstract/document/10223820}
+}
+
+@article{abdi2023cpw,
+  title={CPW-DICE: a novel center and pixel-based weighting for damage segmentation},
+  author={Abdi, Yunus and K{\"u}ll{\"u}, {\"O}mer and Kele{\c{s}}, Mehmet K{\i}v{\i}lc{\i}m and G{\"o}kberk, Berk},
+  journal={Connection Science},
+  volume={35},
+  number={1},
+  pages={2259115},
+  year={2023},
+  publisher={Taylor \& Francis},
+  abstract={Reliable evaluation of damage in vehicles is a primary concern in the insurance industry. Consequently, solutions enhanced with Artificial Intelligence (AI) have become the norm. During the assessment, precise damage segmentation plays a crucial role. Dent is a type of damage that can commonly occur in vehicles. It is difficult to pinpoint and tends to blend in with the background. This paper proposes a novel loss function to improve dent segmentation accuracy in vehicle insurance claims. Centre and Pixel-based Weighted DICE (CPW-DICE) is a loss function that performs pixel-based weighting. The CPW-DICE aims to concentrate on the centre of the dent damage to lessen faulty segmentations. CPW-DICE generates a weight mask during training by employing ground truth (GT) and prediction masks. Simultaneously, the weight mask is incorporated into DICE loss. Experiments conducted on our comprehensive internal dataset show a 3% improvement in Intersection over Union (IoU) score for three state-of-the-art (SOTA) approaches compared to DICE loss. Finally, CPW-DICE is evaluated in similar tasks to demonstrate its benefits beyond car damage segmentation.},
+  url={https://www.tandfonline.com/doi/full/10.1080/09540091.2023.2259115}
+}
+
+@inproceedings{erdem2024analysis,
+  title={Analysis of Viewpoint Dependency in Gait Recognition},
+  author={Erdem, Da{\u{g}}lar Berk and G{\"o}kberk, Berk},
+  booktitle={2024 32nd Signal Processing and Communications Applications Conference (SIU)},
+  pages={1--4},
+  year={2024},
+  organization={IEEE},
+  abstract={This study investigates the viewpoint dependency in gait recognition methods. Gait recognition is the process of identifying individuals based on their walking patterns. The performance of gait recognition systems is significantly affected by the viewpoint of the camera relative to the subject. Our research proposes a new evaluation strategy to understand the effect of 3D gait information in viewpoint dependency under varying viewpoint conditions. Specifically, we utilize the best performing models models within our proposed protocol to explore their performance in addressing viewpoint variations. Our findings reveal that, there is a need for better gait recognition models that can respond more effectively to the changing viewpoint.},
+  url={https://ieeexplore.ieee.org/abstract/document/10600902}
+}
+
+@inproceedings{ozccelik2024self,
+  title={Self-Supervised Dense Visual Representation Learning},
+  author={{\"O}z{\c{c}}el{\.I}k, Timoteos Onur and G{\"o}kberk, Berk and Akarun, Lale},
+  booktitle={2024 32nd Signal Processing and Communications Applications Conference (SIU)},
+  pages={1--4},
+  year={2024},
+  organization={IEEE},
+  abstract={Self-supervised representation learning has shown promising results in recent years. However, most of the proposed methods are pre-trained on object-centric datasets with image-level pretext tasks. In this study, we follow DenseCL, which is pre-trained on pixel-level scene-centric datasets with contrastive learning. Our goal is to alleviate the false negative pairing problem in contrastive learning by consistency regularization. Our method outperforms DenseCL and PixContrast models in most of the scenarios. In PASCAL VOC object detection, we see 0.2% AP50 and 0.3% AP improvements. In COCO object detection, we get 0.3% AP and 0.7% AP boosts. We also improve by 0.4% AP and 0.6% AP in COCO instance segmentation, and 0.1% mAP and 0.9% mAP in PASCAL VOC semantic segmentation. Moreover, attention map visualization and k-nearest neighbour retrieval indicate qualitative improvement from the proposed method.},
+  url={https://ieeexplore.ieee.org/abstract/document/10600771}
+}
+
+@inproceedings{kantarci2024novel,
+  title={A novel part-based benchmark for 3D object reconstruction},
+  author={Kantarc{\i}, Merve G{\"u}l and G{\"o}kberk, Berk and Akarun, Lale},
+  booktitle={2024 32nd Signal Processing and Communications Applications Conference (SIU)},
+  pages={1--4},
+  year={2024},
+  organization={IEEE},
+  abstract={Numerous deep learning-based methods have been proposed for achieving high accuracy in 3D object reconstruction. However, when examining the recent models, we observe that their performances are very close. We claim that more detailed evaluation methods are needed to broaden the comparisons and allow new research directions. Accordingly, in this study, we propose a novel benchmark to evaluate at the part level over three state-of-the-art reconstruction models using the novel rich dataset, 3DCoMPaT++. To evaluate holistic shape reconstruction outputs at the part level, the Part F-Score metric is proposed. Adapting a dataset proposed from a close domain is important for enabling new data to 3D object reconstruction applications and for guiding new adaptations.},
+  url={https://ieeexplore.ieee.org/abstract/document/10600720}  
+}
+
 @inproceedings{girgin2024detection,
   title={Detection and quantification of occlusion for 3d human pose and shape estimation},
   author={Girgin, Emre and G{\"o}kberk, Berk and Akarun, Lale},