Temporal Grounding on ActivityNet-CON
38.8Grounding Success RateTimeChat
Evaluation Results
| Method | Links | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TimeChatData=VTune, FT=TCAS2025.10 | 38.8 | 31 | — | 11.1 | — | 20.8 | — | 21.5 | — | — | — | — | — | |
| TimeChatData=TimeIT, FT=TCAS2025.10 | 38.1 | 29.2 | — | 10.4 | — | 24.2 | — | 18.3 | — | — | — | — | — | |
| Gemini 1.5 Flash#Frames=1 fps2024.11 | 37.8 | 32.4 | 85.7 | 2,710 | 71.7 | 24.8 | 65.8 | 26.8 | 70.8 | — | — | — | — | |
| gemini-1.5tuning=none2025.10 | 37.8 | — | — | — | — | — | — | — | — | 30.8 | 24.8 | 22.4 | 26.8 | |
| TimeChatData=VTune, FT=SFT2025.10 | 37.4 | 28.3 | — | 10.6 | — | 19.6 | — | 19.3 | — | — | — | — | — | |
| videollamaData=VTune, FT=TCAS2025.10 | 33.8 | 25.2 | — | 11.1 | — | 20.9 | — | 18.6 | — | — | — | — | — | |
| videollamaData=VTune, FT=SFT2025.10 | 33 | 24.7 | — | 10 | — | 20.2 | — | 17.7 | — | — | — | — | — | |
| VTimeLLMLLM=Vicuna-7B, #Frames=1002024.11 | 31.9 | 26.4 | 82.7 | 890 | 27.9 | 14.1 | 44.2 | 15.8 | 49.5 | — | — | — | — | |
| GPT-4o#Frames=102024.11 | 26.8 | 18.1 | 67.5 | 1,040 | 38.8 | 16.5 | 61.7 | 18.4 | 68.8 | — | — | — | — | |
| GPT-4otuning=none2025.10 | 26.8 | — | — | — | — | — | — | — | — | 18.1 | 10.4 | 16.5 | 18.4 | |
| TimeChatData=TimeIT, FT=SFT2025.10 | 25.3 | 20.2 | — | 7.5 | — | 8.7 | — | 12.6 | — | — | — | — | — | |
| Qwen2.5vlData=VTune, FT=TCAS2025.10 | 18.7 | 15.1 | — | 2.5 | — | 12.3 | — | 14 | — | — | — | — | — | |
| Qwen2.5vlData=VTune, FT=SFT2025.10 | 18.2 | 14.5 | — | 2.5 | — | 12 | — | 13.2 | — | — | — | — | — | |
| Video-LLaVALLM=Vicuna-7B, #Frames=82024.11 | 13.4 | 10 | 74.5 | 310 | 23 | 6.3 | 46.9 | 7 | 52.3 | — | — | — | — | |
| Video-LLaMALLM=Vicuna-7B, #Frames=82024.11 | 12.8 | 8.5 | 66.8 | 720 | 56.8 | 7.3 | 57.5 | 7.5 | 58.9 | — | — | — | — | |
| VideoChat2LLM=Vicuna-7B, #Frames=162024.11 | 10.5 | 8.7 | 82.8 | 60 | 6 | 5.6 | 53.3 | 5.4 | 51.9 | — | — | — | — | |
| Video-LLaMA2LLM=Mistral-7B, #Frames=82024.11 | 10.4 | 8.2 | 78.6 | 150 | 14.8 | 5.4 | 52.4 | 5.7 | 54.7 | — | — | — | — | |
| VTG-LLMLLM=Llama2-7B, #Frames=962024.11 | 6.8 | 5.3 | 78 | 20 | 3 | 0.7 | 10.9 | 1.7 | 24.9 | — | — | — | — | |
| TimeChatLLM=Llama2-7B, #Frames=962024.11 | 4.6 | 2.9 | 64.1 | 100 | 21.2 | 2.1 | 46.7 | 2.4 | 52.2 | — | — | — | — | |
| TimeChatData=✗, FT=✗2025.10 | 4.6 | 2.9 | — | 1 | — | 2.1 | — | 2.4 | — | — | — | — | — | |
| Video-ChatGPTLLM=Vicuna-7B, #Frames=1002024.11 | 3.3 | 2.8 | 84 | 10 | 4.1 | 1.7 | 50.4 | 1.6 | 49.2 | — | — | — | — |