DSC180F20W4

Participation Assignment (due October 27th at noon)

Watch the video and read the detailed explanation of the COCO dataset

Answer these questions:

• What does your COCO demo do differently than the default demo?
• Show a sample output of your demo (can be a mock-up of what you want it to be)

Replication Checkpoint 1 (due October 30th at 11:59pm)

Report Portion

Write an introduction including:

• An explanation of the problem being investigated: how can an image classification algorithm be trusted to give the correct results?
• A brief explanation of the context of the problem and why it’s interesting.
• A description of the type of data for which the method is appropriate: basic description of observed data used in the investigation (COCO dataset) and why it’s appropriate for addressing the problem.

Code Portion

• Modify the COCO demo to create your own that is different than the default demo.
• Your code should be turned in via GitHub. It should:
  • conform to the template structure discussed in lecture,
  • contain a rudimentary data ingestion pipeline,
  • include documentation both in your README.md, describing the purpose of the code, its contents, and how to run it.
  • be runnable via the command python run.py data. Include a data-params.json file in the config directory, which specifies any data-input locations.
  • include the COCO dataset location in your data-params.json

For both the Report and Code Portion, write in a Canvas comment listing what tasks each group member was responsible for.