Update README.md

README.md

@@ -6,7 +6,8 @@ license: openrail
 
 ## Dataset Description
 
-OVDEval is a new benchmark for OVD model, which includes 9 sub-tasks and introduces evaluations on commonsense knowledge, attribute understanding, position understanding, object relation comprehension, and more.
+OVDEval is a new benchmark for OVD model, which includes 9 sub-tasks and introduces evaluations on commonsense knowledge, attribute understanding, position understanding, object relation comprehension, and more. The dataset is meticulously created to provide hard negatives that challenge models' true understanding of visual and linguistic input. Additionally, we identify a problem with the popular Average Precision (AP) metric when benchmarking models on these fine-grained label datasets and propose a new metric called Non-Maximum Suppression Average Precision (NMS-AP) to address this issue.
+
 
 ## Languages
 
@@ -43,10 +44,10 @@ The dataset contains questions in English and code solutions in Python.
       "height": 254,
       "width": 340,
       "text": [
-        "computer without screen on"
+        "computer without screen on"  # "text" represents the positive sample in this image.
       ],
       "neg_text": [
-        "computer with screen on"
+        "computer with screen on" # "neg_text" represents the category for which this image does not belong.
       ]
     }]
 }