Method Drift›Parameter-efficient fine-tuning (LoRA family)
Superseded baseline#126 of 1,113 most-superseded
TimeSformer
Parameter-efficient fine-tuning (LoRA family)
superseded — cited as a baseline and beaten by newer methods
0 papers critique it · 2 beat it on benchmarks
Beaten on benchmarks
Head-to-head results where a newer method reports beating TimeSformer. Values are copied from the source paper's tables — verify against the cited paper.
- VELoRA: A Low-Rank Adaptation Approach for Efficient RGB-Event based Recognition
VELoRA beats TimeSformer · Precision [PokerEvent benchmark]
57.99 vs 55.69
- Can Local Vision-Language Models improve Activity Recognition over Vision Transformers? -- Case Study on Newborn Resuscitation
FT-C-LoRA beats TimeSformer · mAv [test set]
0.91 vs 0.70
Newer alternatives
Recent methods in the same sub-problem, not yet superseded in the knowledge base.