Improvement accuracy in deep learning: An increasing neurons distance approach with the penalty term of loss function

Hu, Xueheng; Wen, Shuhuan; Lam, Hak‐Keung

doi:10.1016/j.ins.2023.119268

Search citation statements

Order By: Relevance

Paper Sections

Select...

Related Work1

Citation Types

Supporting

Mentioning

Contrasting

Year Published

2024

Publication Types

Select...

Article1

Relationship

Self Cite0

Independent1

Authors

Journals

Cited by 1 publication

(1 citation statement)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In these scenarios, the effective behavioral features in the bounding box are reduced, leading to a decrease in YOLO performance. Previous studies have demonstrated that the design of the IoU loss function can impact the accuracy of object detection [ 24 ]. The IoU loss for bounding box regression exhibits significant sensitivity differences across objects of different scales.…”

Section: Related Workmentioning

confidence: 99%

Occlusion Robust Cognitive Engagement Detection in Real-World Classroom

Xiao,

Xu,

Wei

et al. 2024

Sensors

View full text Add to dashboard Cite

Cognitive engagement involves mental and physical involvement, with observable behaviors as indicators. Automatically measuring cognitive engagement can offer valuable insights for instructors. However, object occlusion, inter-class similarity, and intra-class variance make designing an effective detection method challenging. To deal with these problems, we propose the Object-Enhanced–You Only Look Once version 8 nano (OE-YOLOv8n) model. This model employs the YOLOv8n framework with an improved Inner Minimum Point Distance Intersection over Union (IMPDIoU) Loss to detect cognitive engagement. To evaluate the proposed methodology, we construct a real-world Students’ Cognitive Engagement (SCE) dataset. Extensive experiments on the self-built dataset show the superior performance of the proposed model, which improves the detection performance of the five distinct classes with a precision of 92.5%.

show abstract

Section: Related Workmentioning

confidence: 99%