search

fengyang0317 / unsupervised_captioning

Code for Unsupervised Image Captioning
MIT License
215 stars 51 forks source link

Why the result of object detection of coco dataset has 545 object categories? #17

Closed username123062 closed 4 years ago

username123062 commented 4 years ago

As we all know, coco dataset has 80 object categories. But your code result of "object detection" of coco has 545 object categories. So, can you explain why? Thanks

fengyang0317 commented 4 years ago

I am using the detector trained on OpenImage. https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md