I've been working on my own little arxiv scraper to find me some interesting papers on new datasets. And once in a while there's a cluster with a common theme. This week, it seems to be plants!
A New Dataset and Comparative Study for Aphid Cluster Detection
Aphids are tiny little bugs that can be very bad for your crops, so naturally you might be interested in detecting them. This paper describes a dataset to do just that with over 5447 annotated images.
TomatoDIFF: On–plant Tomato Segmentation with Denoising Diffusion Models
Besides bugs, you can also try and detect crops. This is getting more and more feasible with greenhouses too. So the authors of this paper created a dataset of tomatos as well as benchmark. It turns out diffusion models can beat UNET models these days on these kinds of tasks.
TreeFormer: Transformer-based Tree Counting
But what if you want to count the biggest plants of all? This paper explores methods to count trees. They propose a new method that is benchmarked on two pre-existing tree counting datasets, Jiangsu, and Yosemite, but they also contribute a dataset for based on trees in London.