Description

• INFO: Rebuilding App with Dash, since Streamlit has it´s limitations
• This is a follow up demo of my work done on the "On-Cloud-N: Cloud Cover Detection Challenge"
  • Placement: Top 2% (19/854)
  • Github to Solution: https://github.com/max-schaefer-dev/on-cloud-n-19th-place-solution
• This demo showcases the performance of:
  • Each model on its own
  • Adding TTA
  • Model ensembling
• Online available: Cloud Model Demo
• To-Do:
  • prune and quantized to compress model size & speed up inference

Section 1: Choose sample

• Choose from 5 different samples
• Choose one of the models or choose even an ensemble of model you want to predict with (3 in total)
• Choose how many TTA you want to use (Test-Time Augmentations)

Section 2: Model selection & settings

Section 3: Inference & analytics

• Different metric scores
• Predicted vs true_label comparison
• and more...

References

• https://github.com/fcakyon/streamlit-image-comparison