Human fall direction classification based on deep learning methods and moth flame optimization

Awais Khan

doi:10.1016/j.imavis.2026.106055

Human fall direction classification based on deep learning methods and moth flame optimization

2026-06-07

0

PRE

AI

Muhammad Attique Khan

DOI：10.1016/j.imavis.2026.106055

Original

Original request for help

Save

Abstract

En 中文

• The proposed EfficientNet-b0, DenseNet-201, and ResNet-101 deep learning models were adjusted by incorporating a new fully connected layer (FC), ensuring connectivity with the preceding layers. • The proposed MAKNet-100 a new deep learning model based on bottleneck and self-attention mechanism with hundred (100) hidden layers for fall direction classification. The model consists of minimum and smaller learning parameters. • Features extracted from the modified models were utilized for classification. • Moth Flame Optimization (MFO) was employed to select features from the modified deep learning models, with a focus on comparing computational time.

Human fall direction classification based on deep learning methods and moth flame optimization

Abstract

Journal

Organization

Cited Papers