Locating a Directory after Docker Pull

Question

I'm trying to reproduce results from this repository using docker.

I have completed the pull using

docker pull gasparjan/crnn_ocr:cpu

and run it using

docker run --rm -it -v /home:/data -p 8004:8000 gasparjan/crnn_ocr:cpu

To execute the following command:

python3 predict.py --G 0 --model_path %PATH_TO_MODEL% \
--image_path %PATH_TO_IMAGES% \
--validate --num_instances 512 --max_len 21

I require the PATH_TO_MODEL how do I locate the model.json file that was just pulled?

I've tried using solutions from here but unfortunately I'm unable to locate the pulled directory.

I've also tried downloading the repo from Github and then passing the location of the model.json file and it yields an error

FileNotFoundError: [Errno 2] No such file or directory: '/Users/Luke/Downloads/CRNN-OCR-lite-master/models/OCR_IAM_ver1/model.json'

Please Advise.

I'm using macOS Big Sur.

Answer 1

I am a newbie at this. But I think I have a solution.

I pulled the docker image and it seems like if you replace %PATH_TO_MODEL% with models in your python3 command - it works!

Here is what the final command looks like:

python3 predict.py --G 0 --model_path models \
--image_path %PATH_TO_IMAGES% \
--validate --num_instances 512 --max_len 21

If you do ls once you are inside the docker container, you will be able to see all the directories.

__pycache__  boot  dev  home  lib64  mnt     opt         proc  run   srv  tmp       usr       var
bin          data  etc  lib   media  models  predict.py  root  sbin  sys  train.py  utils.py

I am not sure if this helps, but let me know.

Answer 2

After running this line:

docker run --rm -it -v /home:/data -p 8004:8000 gasparjan/crnn_ocr:cpu

you will be dropped inside the container's command line. You can look around with ls:

➜  sudo docker run --rm -it -v /home:/data -p 8004:8000 gasparjan/crnn_ocr:cpu
root@0e2914bb1aad:/# 
root@0e2914bb1aad:/# ls -la
total 124
drwxr-xr-x   1 root root  4096 Apr 27 13:36 .
drwxr-xr-x   1 root root  4096 Apr 27 13:36 ..
-rwxr-xr-x   1 root root     0 Apr 27 13:36 .dockerenv
drwxr-xr-x   1 root root  4096 May 14  2019 bin
drwxr-xr-x   2 root root  4096 Apr 12  2016 boot
drwxr-xr-x   3 root root  4096 Apr 27 13:35 data
drwxr-xr-x   5 root root   360 Apr 27 13:36 dev
drwxr-xr-x   1 root root  4096 Apr 27 13:36 etc
drwxr-xr-x   2 root root  4096 Apr 12  2016 home
drwxr-xr-x   1 root root  4096 May 14  2019 lib
drwxr-xr-x   1 root root  4096 May 14  2019 lib64
drwxr-xr-x   2 root root  4096 Jan 22  2019 media
drwxr-xr-x   2 root root  4096 Jan 22  2019 mnt
drwxr-xr-x   2 root root  4096 May 14  2019 models
drwxr-xr-x   2 root root  4096 Jan 22  2019 opt
-rw-rw-r--   1 root root  8034 May 11  2019 predict.py
dr-xr-xr-x 378 root root     0 Apr 27 13:36 proc
drwx------   1 root root  4096 May 14  2019 root
drwxr-xr-x   1 root root  4096 May 14  2019 run
drwxr-xr-x   1 root root  4096 Jan 22  2019 sbin
drwxr-xr-x   2 root root  4096 Jan 22  2019 srv
dr-xr-xr-x  13 root root     0 Apr 27 13:37 sys
drwxrwxrwt   1 root root  4096 May 14  2019 tmp
-rw-rw-r--   1 root root  8572 May 10  2019 train.py
drwxr-xr-x   1 root root  4096 Jan 22  2019 usr
-rw-rw-r--   1 root root 23446 May 10  2019 utils.py
drwxr-xr-x   1 root root  4096 Jan 22  2019 var
root@0e2914bb1aad:/# ls -la /models
total 55892
drwxr-xr-x 2 root root     4096 May 14  2019 .
drwxr-xr-x 1 root root     4096 Apr 27 13:36 ..
-rw-r--r-- 1 root root      490 Jan 24  2019 arguments.txt
-rw-r--r-- 1 root root 11482392 Mar 24  2019 checkpoint_weights.h5
-rw-r--r-- 1 root root 34189880 Jan 24  2019 final_model.h5
-rw-r--r-- 1 root root 11482392 Jan 24  2019 final_weights.h5
-rw-r--r-- 1 root root      136 Jan 24  2019 loss_history.pickle.dat
-rw-r--r-- 1 root root    31729 Jan 24  2019 model.json
-rw-r--r-- 1 root root    15524 Jan 24  2019 model_summary.txt

So the file is at /models/model.json inside the container!

As for the images, I couldn't find any images inside the container. I think you are meant to supply your own. In that case, look at the command (/home:/data). This is a bind mount which means that /home on your PC and /data inside the container are the same folder now.

So I would try adding some images to your PC's /home folder and then, inside the container, running:

python3 predict.py --G 0 --model_path /models/model.json \
--image_path /data \
--validate --num_instances 512 --max_len 21