README for the equation detection module

Running the equation detection module

1. Clones the AutoMATES repo

2. Clone the matterport Mask-RCNN repo inside the detection subdirectory

   cd automates/equation_extraction/detection
   git clone https://github.com/matterport/Mask_RCNN.git
   

3. Download and extract data

   • Contact us for the data tar file, we don’t have rights to post…

     tar -zxvf output_objdet.tar.gz
     

4. Build the docker container

    docker build -t maskrcnn-gpu .
   

5. Run the docker container using the desired mode (train or detect) and base model (i.e., imagenet)

   • train:

     sudo ./docker.sh python3 equation.py train --dataset=dataset --subset=train --weights=imagenet
     

   • detect (i.e., find AABBs for the test images):

     sudo ./docker.sh python3 equation.py detect --dataset=dataset --subset=test --weights=last
     

     Note: updates with corresponding timestamps will be added with more information regarding the detection model training and evaluation.