柠檬质量控制数据集.zip

 # Lemons quality control dataset :lemon: [](https://zenodo.org/badge/latestdoi/283124758) [](https://opensource.org/licenses/MIT) [](https://semver.org) Lemon dataset has been prepared to investigate the possibilities to tackle the issue of fruit quality control. It contains 2690 annotated images (1056 x 1056 pixels). Raw lemon images have been captured using the procedure described in the following [blogpost](https://blog.softwaremill.com/when-life-gives-you-lemons-create-a-dataset-70522d6b1aa0) and manually annotated using [CVAT](https://github.com/opencv/cvat). Here's an example of raw unannotated data:  and some annotated samples:  ## Labels | Name | Attribute | Type | Default | Description | Example | | --- | --- | --- | --- | --- | --- | | condition | healthy | boolean | true | Determine whether the fruit is healthy. If not regions with identified issues are annotated |  | | condition | greening | boolean | false | Determine whether the fruit contains areas that are not uniformly yellow and have green areas. |  | | image_quality | blurry | boolean | false | Fruit image is blurry |  | | image_quality | cropped | boolean | false | Not all fruit parts are on the image |  | | image_quality | unnatural_color | boolean | false | There are issues with color representation. |  | | image_quality | no_data | boolean | false | There are black spots on the fruit image that do not contain data. |  | | illness | - | region | - | - |    | | gangrene | - | region | - | - |  | | mould | - | region | - | - |  | | blemish | artificial | region + boolean | - | - |   | | dark_style_remains | - | region | - | After pollination the remains of style are preserved in the fruit. A dark area around the remain of style indicates an unhealthy fruit. This place is the region from which the fruit starts rotting or catches mould. |   | | pedicel | - | region | - | Pedicel refers to a structure connecting a single flower to its inflorescence. |  | | artifact | - | region | - | Image contains artifacts i.e. regions that are not related to a fruit and are a result of wrong image processing. Those regions should be identified and described. |  | ## File name You will notice that file names are composed to form a specific identifier e.g.: `0037_G_I_120_A`: `0037` (individual fruit instance), `120` (relative photo angle), `A` (photo position). Some of them are restricted to the original project and cannot be published. ## Download data | File | Format | Version | | --- | --- | --- | | [Lemon Dataset](data/lemon-dataset.zip) | COCO | v1 | [COCO API](https://github.com/cocodataset/cocoapi) can be utilized to read the dataset. ```python from pycocotools.coco import COCO coco = COCO('../lemon-dataset/annotations/instances_default.json') ``` ## Citing If you use the lemons data set in a scientific publication, we would appreciate references to the following paper: Biblatex entry: ```latex @misc{softwaremill_2020, author = {Maciej Adamiak}, title = {Lemons quality control dataset}, institution = {SoftwareMill}, month = jul, year = 2020, doi