Discussion for Table Detection using Deep Learning

Please comment your query related to this repo:
https://github.com/interviewBubble/Table-Detection-using-Deep-Learning

I followed all steps you have mentioned for table detection.
Here is the config I used, the tf records were generated by converting from csv file ('image_id,xmin,ymin,xmax,ymax,label' format) using the dataset transform command.

train:
  # Name used to identify the run. Data inside `job_dir` will be stored under
  # `run_name`.
  run_name: table-detection
  # Base directory in which model checkpoints & summaries (for Tensorboard) will
  # be saved.
  job_dir: jobs/
  save_checkpoint_secs: 10
  save_summaries_secs: 10

  # Number of epochs (complete dataset batches) to run.
  num_epochs: 10

dataset:
  type: object_detection
  # From which directory to read the dataset.
  dir: model/tfdata
  image_preprocessing:
    min_size: 600
    max_size: 1024
  data_augmentation:
    - flip:
        left_right: True
        up_down: True
        prob: 0.5

model:
  type: fasterrcnn
  network:
    # Total number of classes to predict.
    num_classes: 1

I created a checkpoint 93e6875c68d6 but when I tried this command to make a prediction

lumi predict --checkpoint 93e6875c68d6 data/val/9541_023.png

it returned an empty prediction array:

{"file": "data/val/9541_023.png", "objects": []}

Feels like something obvious I might have missed in setting up the config or the number of images to train is low?
Any suggestions?

How to retrain the model?