I am trying to understand how many images with object-detection bounding boxes are in ImageNet dataset, and how many objects are there in these images.
In MS-COCO paper, it says:
"Recently, a detection challenge has been created from 200 object categories using a subset of 400,000 images from ImageNet. An impressive 350,000 objects have been labeled using bounding boxes."
Are there less objects than images? Later in the paper it says:
MS COCO contains considerably more object instances per image (7.7) as compared to ImageNet (3.0) and PASCAL (2.3)
In the ImageNet website it says 1,034,908 images with bounding boxes.
These numbers seem to conflict.
What am I missing? Does anyone know the correct image and object numbers?
Note: I'm not looking for the number of images with classification annotations, just the object detection annotations.