List your company

Object Removal on COCO subset (test)

47S2 ScoreVISTA
3.3214.662637.34May 24, 2026

Evaluation results

MethodLinks
VISTAVisual Encoder=DINOv2, LLM Backbone=Gemma-2-2B-it2026.05474941.43
VisionSAEVisual Encoder=CLIP Local2026.052423530.71
VISTAVisual Encoder=CLIP - Gemma 2B, LLM Backbone=Gemma-2-2B-it2026.05184780.4
Ablation (No SAE)Visual Encoder=CLIP2026.051320670.46
VisionSAEVisual Encoder=CLIP Global2026.05118451.06
Ablation (No SAE)Visual Encoder=DINOv22026.05102880.22
Ablation (No SAE)Visual Encoder=I-JEPA2026.05818740.34
VISTAVisual Encoder=I-JEPA - Gemma 2B, LLM Backbone=Gemma-2-2B-it2026.05825670.41
PatchSAEVisual Encoder=CLIP Local2026.05555400.65