Data for: Unlocking the Power of AI for Fruit Phenotyping: A Genetic Validation Study in Arabidopsis
Authors/Creators
-
1.
Aberystwyth University
- 2. Corporacion Colombiana de Investigación Agropecuaria
Description
This dataset contains annotated images of mature inflorescences, collected from experiments conducted on the Multiparent Advanced Generation Inter-Cross (MAGIC) population. The images were utilised to develop and validate a deep learning-based pipeline for Arabidopsis fruit trait extraction.
- The images and annotations have been separated into a training and a test set.
- The pretrained Cascade Mask-RCNN model (in PyTorch) for instance segmentation of Arabidopsis siliques is also provided in arabidopsis.pth file.
The instructions for installing and testing the pipeline, along with the extracted phenotype data and MAGIC genomic data for QTL analysis, can be found in https://github.com/kieranatkins/silique-detector/ .
For the verification of the pipeline using QTL analysis, the full dataset collection (in total of over 7000 images) has been utilised, and the raw images and metadata are available at the following links.:
AT023: 10.5281/zenodo.13853394
Notes
- The annotations are stored in two JSON files:
at025_v2_fixed_train.json
at025_v2_fixed_test.json
These files contain image annotations for instance segmentation of Arabidopsis siliques in COCO format. They were used for training and testing deep learning models. The training set includes 44 images with 2,288 annotated siliques, while the test set contains 11 images with 652 silique segmentations.
- A list of known genes relevant to silique development is provided in the file:
filtered_genes.csv
They were retrieved from the TAIR9 database, if they have associated gene ontology terms related to fruit development, organ development, organ morphogenesis, or seed development.
Files
images_train.zip
Files
(1.0 GB)
| Name | Size | Download all |
|---|---|---|
|
md5:0e4894c3a067f6bc11405c9a69efbd47
|
490.3 MB | Download |
|
md5:be46dd1f1a5ac0b15e7ecdedf81e700b
|
2.8 MB | Preview Download |
|
md5:68d2dc682c34511bf8fa6f9ab7b5b446
|
9.7 MB | Preview Download |
|
md5:4ab07681a83d77a676433c16fdbb20a2
|
249.1 kB | Preview Download |
|
md5:d73c929437c38d2fbd991c2b1e9558b3
|
128.2 MB | Preview Download |
|
md5:6ee5c82ca2cc588f3b3502f06d48f59b
|
404.3 MB | Preview Download |
Additional details
Software
- Repository URL
- https://github.com/kieranatkins/silique-detector/
- Programming language
- Python , R
- Development Status
- Active