---
license: mit
language:
- en
tags:
- biology
- CV
- images
- animals
- lepidoptera
- butterflies
- segmentation
- heliconius
- forewings
- hindwings
- separated wings
- full body
- butterfly
- RGB
- ruler
- whitebalance
- label
- colorchecker
---

## Model Card for butterfly_segmentation_yolo_v8

This model takes in an image of a butterfly (with or without body attached to wings) and segments out any existing hindwings and forewings, in addition to pictured equipment described below. 

## Model Details

yolov8m_shear_10.0_scale_0.5_translate_0.1_fliplr_0.0_best.pt is the butterfly segmentation model. 

The butterfly segmentation model was trained on a dataset of 800 total images from the Jiggins, OM_STRI, and Monteiro datasets. The model architecture is based on YOLO v8 (yolov8m-seg.pt), which we fine-tune further on our dataset of 800 images. 

## Model Description

The model is responsible for taking an input image (RGB) and generating segmentation masks for all classes below that are found in the image. Data augmentations applied during training include shear (10.0), scale (0.5), and translate (0.1). The model was trained for 50 epochs with an image size of 256. Note that despite defining an image size of 256, the normalized masks predicted by yolo can be rescaled to the original image size. 

### Segmentation Classes
[`pixel class`] corresponding category

- [0] background
- [1] right_forewing
- [2] left_forewing
- [3] right_hindwing
- [4] left_hindwing
- [5] ruler
- [6] white_balance
- [7] label
- [8] color_card
- [9] body

### Details

model.train(data=YAML, 
            imgsz=256,
            epochs=50, 
            batch=16,
            device=DEVICE,
            optimizer='auto',
            verbose=True,
            val=True,
            shear=10.0,
            scale=0.5, 
            translate=0.1,
            fliplr = 0.0 
            )

## Metrics

                Class     Images  Instances      mAP50-95
                   all         64        358      
            background         64          3      0.20946
        right_forewing         64         58      0.9845
         left_forewing         64         51      0.9682 
        right_hindwing         64         59      0.95296
         left_hindwing         64         50      0.93961 
                 ruler         64         31      0.73608
         white_balance         64         18      0.90686
                 label         64         50      0.80865
            color_card         64         24      0.92653
                  body         64         14      0.78283
                  
**Developed by:** Michelle Ramirez

## How to Get Started with the Model

To view applications of how to load in the model file and predict masks on images, please refer to [this github repository](https://github.com/Imageomics/wing-segmentation)

## Citation

**BibTeX:**

```
@software{Ramirez_Lepidoptera_Wing_Segmentation_2024,
  author = {Ramirez, Michelle},
  doi = {10.5281/zenodo.10869579},
  month = mar,
  title = {{Lepidoptera Wing Segmentation}},
  url = {https://github.com/Imageomics/wing-segmentation},
  version = {1.0.0},
  year = {2024}
}
```

**APA:**

Ramirez, M. (2024). Lepidoptera Wing Segmentation (Version 1.0.0) [Computer software]. https://doi.org/10.5281/zenodo.10869579

## Acknowledgements

The [Imageomics Institute](https://imageomics.org) is funded by the US National Science Foundation's Harnessing the Data Revolution (HDR) program under [Award #2118240](https://www.nsf.gov/awardsearch/showAward?AWD_ID=2118240) (Imageomics: A New Frontier of Biological Information Powered by Knowledge-Guided Machine Learning). Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.