List your company

Scene Captioning on WHOOPS! RGBP seen scenes (test)

80.46CIDErQwen3-VL-8B-Instruct RGBP
-1.564819.730141.02562.3199Apr 1, 2026

Evaluation results

MethodLinks
Qwen3-VL-8B-Instruct RGBPBackbone=Qwen3-VL-8B, Dataset=RGBP2026.0480.4616.07
ProCapBackbone=TinyLlama-1.1B2026.0469.9521.87
ProCapBackbone=OpenLLaMA-3B2026.0469.4622.08
Qwen3-VL-8B-Instruct RGBPBackbone=Qwen3-VL-8B, Dataset=RGBP2026.0437.723.59
ProCapBackbone=Vicuna-1.5-7B2026.0436.5322.7
ProCapBackbone=OPT-2.7B2026.0428.1924.11
ProCapBackbone=Vicuna-1.5-7B2026.0422.336.02
ProCapBackbone=OpenLLaMA-3B2026.0419.375.43
ProCapBackbone=OPT-2.7B2026.0415.044.4
ProCapBackbone=TinyLlama-1.1B2026.0411.883.62
Qwen3-VL-8B-InstructParameters=8B2026.0411.1910.97
Qwen3-VL-4B-InstructParameters=4B2026.0410.3510.91
Qwen3-VL-2B-InstructParameters=2B2026.049.8211.05
FastVLM-7BParameters=7B2026.046.256.77
FastVLM-0.5BParameters=0.5B2026.045.545.85
FastVLM-1.5BParameters=1.5B2026.044.146.09
FastVLM-7BParameters=7B2026.042.2111.98
Qwen3-VL-4B-InstructParameters=4B2026.042.0212.15
Qwen3-VL-2B-InstructParameters=2B2026.042.0112.83
Qwen3-VL-8B-InstructParameters=8B2026.041.9713.07
FastVLM-1.5BParameters=1.5B2026.041.749.92
FastVLM-0.5BParameters=0.5B2026.041.5910.17