no code implementations • 20 Oct 2023 • Francisco Eiras, Kemal Oksuz, Adel Bibi, Philip H. S. Torr, Puneet K. Dokania
Referring Image Segmentation (RIS) - the problem of identifying objects in images through natural language sentences - is a challenging task currently mostly solved through supervised learning.
no code implementations • 26 Sep 2023 • Kemal Oksuz, Selim Kuzucu, Tom Joy, Puneet K. Dokania
Combining the strengths of many existing predictors to obtain a Mixture of Experts which is superior to its individual components is an effective way to improve the performance without having to develop new architectures or train a model from scratch.
Ranked #1 on Oriented Object Detection on DOTA 1.0
1 code implementation • CVPR 2023 • Kemal Oksuz, Tom Joy, Puneet K. Dokania
The current approach for testing the robustness of object detectors suffers from serious deficiencies such as improper methods of performing out-of-distribution detection and using calibration metrics which do not consider both localisation and classification quality.
1 code implementation • 3 Jan 2023 • Fehmi Kahraman, Kemal Oksuz, Sinan Kalkan, Emre Akbas
(ii) Motivated by our observations, e. g., that NMS-free detectors can also benefit from correlation, we propose Correlation Loss, a novel plug-in loss function that improves the performance of various object detectors by directly optimizing correlation coefficients: E. g., Correlation Loss on Sparse R-CNN, an NMS-free method, yields 1. 6 AP gain on COCO and 1. 8 AP gain on Cityscapes dataset.
1 code implementation • 19 Oct 2021 • Kemal Oksuz, Baris Can Cam, Fehmi Kahraman, Zeynep Sonat Baltaci, Sinan Kalkan, Emre Akbas
We present the effectiveness of maIoU on a state-of-the-art (SOTA) assigner, ATSS, by replacing IoU operation by our maIoU and training YOLACT, a SOTA real-time instance segmentation method.
Ranked #9 on Real-time Instance Segmentation on MSCOCO
3 code implementations • ICCV 2021 • Kemal Oksuz, Baris Can Cam, Emre Akbas, Sinan Kalkan
RS Loss supervises the classifier, a sub-network of these methods, to rank each positive above all negatives as well as to sort positives among themselves with respect to (wrt.)
2 code implementations • 21 Nov 2020 • Kemal Oksuz, Baris Can Cam, Sinan Kalkan, Emre Akbas
Despite being widely used as a performance measure for visual detection tasks, Average Precision (AP) is limited in (i) reflecting localisation quality, (ii) interpretability and (iii) robustness to the design choices regarding its computation, and its applicability to outputs without confidence scores.
3 code implementations • NeurIPS 2020 • Kemal Oksuz, Baris Can Cam, Emre Akbas, Sinan Kalkan
We propose average Localisation-Recall-Precision (aLRP), a unified, bounded, balanced and ranking-based loss function for both classification and localisation tasks in object detection.
Ranked #86 on Object Detection on COCO test-dev
1 code implementation • 21 Sep 2019 • Kemal Oksuz, Baris Can Cam, Emre Akbas, Sinan Kalkan
Using our generator as an analysis tool, we show that (i) IoU imbalance has an adverse effect on performance, (ii) hard positive example mining improves the performance only for certain input IoU distributions, and (iii) the imbalance among the foreground classes has an adverse effect on performance and that it can be alleviated at the batch level.
Ranked #194 on Object Detection on COCO minival
1 code implementation • 31 Aug 2019 • Kemal Oksuz, Baris Can Cam, Sinan Kalkan, Emre Akbas
In this paper, we present a comprehensive review of the imbalance problems in object detection.
3 code implementations • ECCV 2018 • Kemal Oksuz, Baris Can Cam, Emre Akbas, Sinan Kalkan
Moreover, we present LRP results of a simple online video object detector which uses a SOTA still image object detector and show that the class-specific optimized thresholds increase the accuracy against the common approach of using a general threshold for all classes.