Population counting

 


This addresses the problem of counting and locating specific objects from images. The shape of the objects is relatively simple in the proposed datasets. The main difficulty comes from the partially occluded objects.


Evaluation of the results: The results can be evaluated according to three criteria: the estimated number of objects, the precision of their position, and time. You can use this Matlab code to evaluate your results from the following datasets.



 

Bird_Colony_small
Sample of an aerial image representing a colony of migrating birds
Acknowledgment: Tour du Valat


Image (160*120 pixels)
Ground truth

Results


 

Bird_Colony
Aerial image representing a colony of migrating birds
Acknowledgment: Tour du Valat


Image (3010*1960 pixels)
Ground truth not available, number of birds estimated by an expert: 10800

Results

 

 

Stomata
Microscope image of leaf in which opened stomata (back dots) must be counted


Image (657*617 pixels)
Ground truth

Results

 

 

Cells
Simulated microscope image of cells
Acknowledgment: [Lehmussola et al.,TMI 2007]


Image (1000*1000 pixels)
Ground truth

Results

 

 

Yellow_cabs
Aerial image of a city


Image (1324*924 pixels)
Ground truth

Results

 


 


back