I have a very limited dataset of around 12k grayscale images and wanted to know if there is a CNN model that I can use for fine tuning or an grayscale image dataset that can be used for pre-training. I dont want to use models trained on ImageNet as I dont want to convert my grayscale images to color images.
There is a very interesting project led by Prof Greg Shakhnarovich, using the task of automatic colorization for pre-training deep nets. You can find on the project's web site pre-trained models for gray-scale images.
AFAIK, there are other research groups working on similar tasks, you might be able to google similar projects/models by other groups.
Alternatively, you can convert imagenet images to gray scale and per-train on gray scale images yourself.
- Solved – Singular value decomposition on RGB images
- Solved – Making a heatmap understandable when printed in black ink
- Solved – How to prepare the image data as a training data for Deep Learning
- Solved – Data Augmentation using Eigenvalues and Eigenvectors
- Solved – How to do weight normalization in VGG network for style transfer