Counting the uncountable: deep semantic density estimation from Space
This addresses the challenge of counting uncountable small objects in remote sensing for applications like urban planning or environmental monitoring, representing an incremental improvement by adapting existing computer vision techniques to a specific domain.
The paper tackles the problem of counting small objects in satellite images where traditional detection is infeasible due to low resolution, by proposing a deep semantic density estimation method that combines density estimation with semantic segmentation in a CNN, achieving robust counting of objects in cluttered scenes.
We propose a new method to count objects of specific categories that are significantly smaller than the ground sampling distance of a satellite image. This task is hard due to the cluttered nature of scenes where different object categories occur. Target objects can be partially occluded, vary in appearance within the same class and look alike to different categories. Since traditional object detection is infeasible due to the small size of objects with respect to the pixel size, we cast object counting as a density estimation problem. To distinguish objects of different classes, our approach combines density estimation with semantic segmentation in an end-to-end learnable convolutional neural network (CNN). Experiments show that deep semantic density estimation can robustly count objects of various classes in cluttered scenes. Experiments also suggest that we need specific CNN architectures in remote sensing instead of blindly applying existing ones from computer vision.