EchoRaven/safewatch-genai

SafeWatch-TV2V Dataset Overview This dataset combines SafeWatch-Bench original videos and TV2V generated videos for video safety detection and robustness testing. Dataset Structure Three Main Components: TV2V-output: Generated videos with prompts Videos: 1,040 files Metadata: 1,213 entries in TV2V-output.jsonl Content: Generated videos with their corresponding prompts TV2V-input: Original videos used for generation Videos: 410… See the full description on the dataset page: https://huggingface.co/datasets/EchoRaven/safewatch-genai.