REVIEW 6 cited by
SIoU Loss: More Powerful Learning for Bounding Box Regression
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
The effectiveness of Object Detection, one of the central problems in computer vision tasks, highly depends on the definition of the loss function - a measure of how accurately your ML model can predict the expected outcome. Conventional object detection loss functions depend on aggregation of metrics of bounding box regression such as the distance, overlap area and aspect ratio of the predicted and ground truth boxes (i.e. GIoU, CIoU, ICIoU etc). However, none of the methods proposed and used to date considers the direction of the mismatch between the desired ground box and the predicted, "experimental" box. This shortage results in slower and less effective convergence as the predicted box can "wander around" during the training process and eventually end up producing a worse model. In this paper a new loss function SIoU was suggested, where penalty metrics were redefined considering the angle of the vector between the desired regression. Applied to conventional Neural Networks and datasets it is shown that SIoU improves both the speed of training and the accuracy of the inference. The effectiveness of the proposed loss function was revealed in a number of simulations and tests.
Forward citations
Cited by 6 Pith papers
-
An Uncertainty-aware DETR Enhancement Framework for Object Detection
Adding a Gaussian-box Gromov-Wasserstein loss and Bayes-risk-based refinement to DETR detectors improves their AP on COCO and leukocyte datasets while producing localization uncertainty estimates.
-
Towards Continuous Home Cage Monitoring: An Evaluation of Tracking and Identification Strategies for Laboratory Mice
A real-time mouse tracking and ear-tag identity pipeline reports 95.28% identification accuracy and fewer ID switches than SLEAP and DeepLabCut on a 100-minute home-cage dataset.
-
RT-VLM: Re-Thinking Vision Language Model with 4-Clues for Real-World Object Recognition Robustness
Fine-tuning a vision-language model on a synthetic 4-clue dataset and adding a self-critique inference loop improves robustness to domain shifts in object recognition.
-
YOLOv14:Unified Cross-Domain Real-Time Object Detectionwith Adaptive Multi-View Representation
YOLOv14 reports a real-time detector with 49.1 COCO mAP and cross-domain gains, but its headline game-character gain is measured on a benchmark synthesized from its own training augmentation.
-
An Intelligent-Cloud Edge Multimodal Interaction System for Robots
A cloud-edge robot system combined a CBAM/DIoU-enhanced YOLO11n gesture detector with LLM/VLM agents, reporting 95–98.9% precision and 82–95% task success on small, validation-based evaluations.
-
VR-YOLO: Enhancing PCB Defect Detection with Viewpoint Robustness Based on YOLO
VR-YOLO, a YOLOv8 variant with SIoU loss, CBAM attention, and augmented training data, reports mAP 98.9% on the PKU PCB dataset and 94.7% on sheared/rotated test images, but the viewpoint evaluation is in-distribution...
Discussion (0). Sign in to comment.