Segmentation of cotyledon

biomag-lab / hypocotyl-UNet

MIT License

11 stars 2 forks source link

Segmentation of cotyledon #7

Open darvida opened 4 years ago

darvida commented 4 years ago

Hello i'm having problems with the segmentation of the cotyledons bot not with the stems. Did you have any problems with this, and if so how did you overcome this? The problem might be that it is hard for the model to distinguish between the stem and cotyledon when I use grayscale images. I have currently trained the model with about 300 images including augmentation. 13 1-70 00 s

cosmic-cortex commented 4 years ago

Hi! Can you send me a ground truth image with annotations? Based on the image you uploaded above, it seems like that there should be no problem with this.

darvida commented 4 years ago

13 5-80 00 s 13 1-90 00 s

Here are two examples of how i'm annotating the images. I'm having the same issue on all of the images that i'm predicting on, but it might have to do with the amount of training data.