Dimensions overall score 7.0

GitHub Code Pulse

No public code linked for this paper yet.

• Current Video Quality Assessment (VQA) predictions largely underrepresent video fluency.result 90%
• Video Fluency Assessment (VFA) is proposed as a standalone perceptual task focused on the temporal dimension of videos.method 95%
• The FluVid dataset was constructed with 4,606 in-the-wild videos, featuring a balanced fluency distribution and the first-ever scoring criteria and human study for VFA.method 95%
• A benchmark of 23 methods was developed on the FluVid dataset, representing the most comprehensive evaluation to date for VFA.method 90%
• FluNet, a baseline model for VFA, incorporates Temporal Permuted Self-Attention (T-PSA) to enhance temporal information processing and inter-frame interactions.method 95%
• The proposed FluNet achieves state-of-the-art performance on the VFA task.result 85%
• The T-PSA module in FluNet allows for increased temporal side length (D) while maintaining computational cost by compressing weight matrices.technical 90%
• A benchmark of 23 methods was developed on the FluVid dataset, representing the most comprehensive evaluation to date for VFA.method 90%
• Current Video Quality Assessment (VQA) predictions largely underrepresent video fluency, limiting their applicability.limitation 90%
• Video Fluency Assessment (VFA) is proposed as a novel, standalone perceptual task focused on the temporal dimension of videos.method 95%
• The FluVid dataset was constructed with 4,606 in-the-wild videos, featuring balanced fluency distribution and the first-ever scoring criteria and human study for VFA.method 95%
• FluNet, a proposed baseline model for VFA, incorporates Temporal Permuted Self-Attention (T-PSA) to enhance temporal information processing and inter-frame interactions.method 95%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card