Researchers have evaluated the effectiveness of different frame rates and neural network architectures for classifying autism-related self-stimulatory behaviors from video. Using the Self-Stimulatory Behavior Diagnosis (SSBD) dataset, Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models achieved peak accuracies of 97.5% and 98.75% respectively, with an optimal sampling interval of every 15 frames. The study also explored data augmentation strategies, finding horizontal flip to be the most effective standalone technique and upsampling to be crucial for complex behavioral video augmentation. These findings offer practical guidance for developing scalable computational methods for remote behavioral screening in clinical settings. AI
IMPACT Provides guidance on optimal frame rates and architectures for video-based behavioral classification in clinical settings.
RANK_REASON The cluster contains an academic paper detailing research findings on AI model performance for a specific classification task.
- ACM Symposium on Interactive 3D Graphics and Games
- arXiv
- autism
- CNN
- cs.CV
- gated recurrent unit
- long short-term memory
- Self-Stimulatory Behavior Diagnosis
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
