Verg-Avesta / CounTR

CounTR: Transformer-based Generalised Visual Counting
https://verg-avesta.github.io/CounTR_Webpage/
MIT License
92 stars 9 forks source link

Counting Multiple object class #13

Closed jaideep11061982 closed 1 year ago

jaideep11061982 commented 1 year ago

If I have got objects having different shapes,sizes and colors then do I need exemplar bounding box for each different type of object varrying in shapes/sizes/colors to count Say I have total 40 objects , 8 Different type of objects having 5 instances each. Do I need min 8 exampler boxes or it can be less object type @WeidiXie @Verg-Avesta

Verg-Avesta commented 1 year ago

The model has generalization ability and is robust to small differences in color, shape, and size. Therefore, it is only needed to select a few representative exemplars for counting.

In fact, objects are generally not considered to belong to the same category when their shapes differ greatly. If you want to count objects with great shape difference, it is better to count them as different categories and each time count one category.

jaideep11061982 commented 1 year ago

@Verg-Avesta thanku for quick response. During inference how can one get the exemplar bounding box. In your demo you have hard coded it for the image. But in real world how those bounding boxes can be generated and passed to the model

Verg-Avesta commented 1 year ago

In real world, the user should also draw several bound boxes in the image to clarify which category the user wants to count. This is what have to be done by the users because if there are multiple classes of objects, no one will know which class to count except the user.