Update README.md

README.md

@@ -160,7 +160,7 @@ size_categories:
 - ⚖️ **Splits:**
   - Validation split (~50%) for experimentation and model fine-tuning.
   - Test split (~50%) with hidden ground-truths for public leaderboard evaluation.
-- 🔎 **Annotation Quality:** All images and traces manually collected and labeled by human experts.
+- 🔎 **Annotation Quality:** All images are manually collected, labeled with instructions as well as traces, and reviewed by human experts.
 - 🏅 **Evaluation Metric:** Semantic-aware Trace Score, combining Dynamic Time Warping distance, goal endpoint error, and embodiment-conditioned semantic penalties.
 
 ## Uses