Training with grayscale images

experiencor / keras-yolo2

Easy training on custom dataset. Various backends (MobileNet and SqueezeNet) supported. A YOLO demo to detect raccoon run entirely in brower is accessible at https://git.io/vF7vI (not on Windows).

MIT License

1.73k stars 785 forks source link

Training with grayscale images #351

Open ghffarpk opened 6 years ago

ghffarpk commented 6 years ago

I want to train the model on grayscale images. Is it only required to change line input_image = Input(shape=(IMAGE_H, IMAGE_W, 3)) to input_image = Input(shape=(IMAGE_H, IMAGE_W, 1)) in step by step guide given in code ?

rodrigo2019 commented 6 years ago

you can use my fork, training with grayscale images is already avaliable

ghffarpk commented 6 years ago

@rodrigo2019 i am using the code you provided. should i provide pre trained model for it ? also how much time it will take ?