search

baomingwang / MTCNN-Tensorflow

mtcnn-tf
MIT License
162 stars 96 forks source link

all-in-one model #4

Open waquey opened 6 years ago

waquey commented 6 years ago

Hello, I've tried to reproduce the training process based on your description. Currently, I can generate P/R/O checkpoint respectively.

However, could you please help explain how we can combine these 3 ckpt(or pb) into the model you've provided in all-in-one model?

Thanks

liangheng commented 6 years ago

I have the same question. Can you answer it

ankur219 commented 5 years ago

Did anyone get a solution to this problem?

ankur219 commented 5 years ago

Well, there is no need to get a combined model. Inside the model directory, you'll have to create three folders of the name 'pnet', 'rnet' and 'onet' and store inside them your checkpoints and meta data files.

yeyupiaoling commented 5 years ago

@ankur219 But test need a model to predict. So we should combine model.

seanxu889 commented 5 years ago

@ankur219 It said we should run python test_img.py YOUR_IMAGE_PATH --model_dir ./save_model/all_in_one/ to test mtcnn. However, after generating P/R/O model respectively. How can we combine them to an all_in_one file?

sondv2 commented 5 years ago

But how to create single file model ?

bobinush commented 4 years ago

Sigh, the author left out the most important thing which is to combine these..?