CarrotWeeds: An Annotated Image Dataset for Weed Detection in Carrot Cultivation

CarrotWeeds: An Annotated Image Dataset for Weed Detection in Carrot Cultivation This dataset contains annotated images collected in the context of carrot cultivation. The annotations cover both carrot plants and associated weed species. Each plant is annotated with a bounding box and a keypoint indicating the location of its apical meristem. - Version: 1.0.0- Annotation format: Pascal VOC XML with keypoints- Background representation: empty XML without <object>- Class abbreviations are explained in metadata/classes.csv- Further metadata can be found in metadata/metadata.yaml Dataset summary Category Images Annotations Matched pairs Instances labeled 3266 3266 3266 19339 background 855 855 855 0 TOTAL 4121 4121 4121 19339 Instances per class Background is an annotation state and is not included as an object class. Class Instances Percent 1TRFG 164 0.85% AMACH 115 0.59% AMARE 21 0.11% CAPBP 70 0.36% CHEAL 1450 7.50% DAUCS 7463 38.59% EROCI 2 0.01% GASPA 11 0.06% GERMO 7 0.04% LACSE 1 0.01% LAMAL 73 0.38% POLAV 109 0.56% POROL 10 0.05% SENJA 6 0.03% SINAR 70 0.36% SONAR 5 0.03% SPRAR 726 3.75% STEME 7 0.04% THLAR 5 0.03% grasses 651 3.37% weed_cot 8137 42.08% weed_dicotyl 2 0.01% weed_other 234 1.21% TOTAL 19339 100.00% Directory semantics - data/images/labeled and data/annotations/labeled/ contain paired samples.- data/images/background and data/annotations/background contain paired soil-background images without annotated plant instances.- Background XMLs are present but contain no <object> element. License This dataset is licensed under the Creative Commons Attribution 4.0International License (https://creativecommons.org/licenses/by/4.0/) (CC BY 4.0)