Overview

Dataset stats

• 27,574개의 이미지가 존재
  • 25,574 for training
  • 2000 for testing
• 3,688개 category의 707,868개 Object존재
• 193,238 개의 Object part (그리고 parts of parts)가 annote되어있음
• Polygon annotation 존재

index_ade20k.pkl

dataset의 각 파일에 대한 정보가 담겨져 있다. 아래는 description이다.

• filename array of length N=27574 with the image file names
• folder array of length N with the image folder names.
• scene array of length N providing the scene name (same classes as the Places database) for each image.
• objectIsPart array of size [C, N] counting how many times an object is a part in each image. objectIsPart[c,i]=m if in image i object class c is a part of another object m times. For objects, objectIsPart[c,i]=0, and for parts we will find: objectIsPart[c,i] = objectPresence(c,i)
• objectPresence array of size [C, N] with the object counts per image. objectPresence(c,i)=n if in image i there are n instances of object class c.
• objectcounts array of length C with the number of instances for each object class.
• objectnames array of length C with the object class names.
• proportionClassIsPart array of length C with the proportion of times that class c behaves as a part. If proportionClassIsPart[c]=0 then it means that this is a main object (e.g., car, chair, ...). See bellow for a discussion on the utility of this variable.
• wordnet_found array of length C. It indicates if the objectname was found in Wordnet.