Add the OBJECT_DETECTION task

[nudles]

For this task, the input data (for inference) would be an image, and the output data should be a list
like of tuples: (bounding box, probability of each category).
Pls refer to https://github.com/shwars/mPyPl/wiki/Reading-PASCAL-VOC-Format

Discussion is welcomed.

[pinpom]

I found this colab on using Mask-RCNN for detecting Pneumonia from DICOM format images. Input is image + annotations .csv file and Output is .csv file (patientID ; bounding box string -- x, y, width, height)
https://colab.research.google.com/github/mdai/ml-lessons/blob/master/lesson3-rsna-pneumonia-detection-kaggle.ipynb?

Kaggle dataset: https://www.kaggle.com/c/rsna-pneumonia-detection-challenge/data

[chrishkchris]

A web UI for object detection has been added to the application zoo in #31