Video-Text-to-Text
PEFT
Safetensors
English
ai-generated-video-detection
video-forensics
omni-modal
audio-visual
agent
multi-agent
tool-use
reinforcement-learning
lora
qwen3.5
Instructions to use tencent/WeVisForensics with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use tencent/WeVisForensics with PEFT:
Task type is invalid.
- Notebooks
- Google Colab
- Kaggle
File size: 1,431 Bytes
9fa025a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 | {
"roles": {
"physical_consistency": [],
"visual_appearance": [
{
"id": "R1",
"text": "Ignore: Motion blur that softens a moving subject's skin or hair boundaries during fast movement."
},
{
"id": "R2",
"text": "Report: A subject's surface lacking fine texture while adjacent high-frequency details remain sharp."
},
{
"id": "R3",
"text": "Ignore: Global softness or compression artifacts that uniformly reduce detail across the entire frame."
},
{
"id": "R4",
"text": "Ignore: Smooth skin or facial surfaces that lack fine pore detail due to beauty filters or high compression."
},
{
"id": "R5",
"text": "Report: Facial or hand contours that locally melt or duplicate while surrounding background details remain sharp."
}
],
"common_sense": [
{
"id": "R1",
"text": "Report: A character's body or props shifting into a physically incompatible form or scale during a transition."
},
{
"id": "R2",
"text": "Report: A physical object appearing on the ground with a spill or impact mark lacking a visible causal action."
},
{
"id": "R3",
"text": "Ignore: Stylized or game-like environments where objects or animals behave according to internal rules."
}
],
"ocr_consistency": []
}
}
|