Project: Classification of Mixed CIFAR-10 Images

Description

The model takes as input an image obtained as the average of two randomly selected samples from the CIFAR-10 dataset and must predict the categories of the two original components.

Dataset Details

• The first image belongs to one of the following 5 categories:
  • Airplane
  • Automobile
  • Bird
  • Cat
  • Deer
• The second image belongs to one of the remaining 5 categories:
  • Dog
  • Frog
  • Horse
  • Ship
  • Truck

The model must output two labels, each belonging to one of the two defined groups (each with a range of 5 values).

Evaluation Methodology

The model's performance will be measured using the following metric:

1. Compute accuracy for classifying both image components.
2. Calculate the average of the two obtained accuracies.

Testing and Validation

• The metric must be evaluated on 10,000 inputs generated from test data.
• The calculation must be repeated 10 times to estimate the standard deviation.
• The standard deviation value must be reported.

Input Data and Examples

A data generator will be provided below, along with some examples to support understanding of the task.

Performance

The proposed model achieves an accuracy of 0.827