From the model card
Fields below are copied from the tags and counters on the HuggingFace repository KingTechnician/videomae-small-finetuned-kinetics-xd-violence-binary at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- KingTechnician
- Pipeline tag
- video-classification
- Library
- Transformers
- Weight formats
- safetensors
- License tag
cc-by-nc-4.0— read the license file in the repo before relying on it- Lineage
-
- base model MCG-NJU/videomae-small-finetuned-kinetics
- fine-tune of MCG-NJU/videomae-small-finetuned-kinetics
- Downloads (HF counter at last fetch)
- 391,007
- Likes (HF counter at last fetch)
- 0
- Model card
- https://huggingface.co/KingTechnician/videomae-small-finetuned-kinetics-xd-violence-binary
Use cases
- Automated violence detection in user-generated video content
- Content moderation for video platforms before manual review
- Pre-filtering video training datasets to remove violent content
- Safety screening for video surveillance or broadcast content
- Research baseline for temporal violence detection models
Pros
- VideoMAE pre-training provides strong temporal features with limited labelled data
- XD-Violence covers diverse violence scenarios beyond typical action recognition datasets
- HuggingFace endpoints compatible; straightforward integration
Cons
- Binary only; no nuanced violence category classification
- 0 community likes; no independent published evaluation of this specific checkpoint
- VideoMAE-Small may miss subtle violence cues requiring larger temporal context
- No license specified; verify before commercial content moderation deployment
Tags
transformerssafetensorsvideomaevideo-classificationgenerated_from_trainerbase_model:MCG-NJU/videomae-small-finetuned-kineticsbase_model:finetune:MCG-NJU/videomae-small-finetuned-kineticslicense:cc-by-nc-4.0endpoints_compatibleregion:us