Computer Vision Challenge: Identify Missing Annotations

Scenario: My chocolate detection model is struggling to identify a particular class: Coconut. There may be a few missing annotations in my dataset for the Coconut chocolate. It is essential that the model identifies chocolates as accurately as possible. A mistake could lead to some unhappy customers.

The challenge: Write a program that identifies all images with missing annotations in a chocolate dataset.

Here is an example of an image in the dataset that has been annotated with all chocolates:

The expected deliverables for this challenge are:

1. A Python script that you used to find the images with missing annotations, and;
2. The names of the files with missing annotations.

Here is the chocolate dataset, formatted in the COCO JSON format. The dataset is distributed as a ZIP file.

If you complete this challenge, send an email to readers [at] jamesg [dot] blog and tell me how you solved it. If you enjoyed completing the challenge, Roboflow is hiring.