supervisely-ecosystem / seeds

Demo project with photos of seeds

Seeds

Overview • Statistics • Examples •

Overview

Demo project with seeds photos for research. Can be used as a playground for:

• testing different models: detection / segmentation / instance segmentation
• demo for synthetic data
• all images have the same resolution: 4032 * 2268 => so it is a good test for sliding window approaches
• modeling some close tasks from different industries, images may have similar properties in terms of lighting conditions, number of objects and their sizes, camera positions, etc; for example:
  • agriculture (find plants diseases, count items, inspect their properties (e.g. size) and quality)
  • medicine (x-ray)
  • microbiology (microscope images)
  • visual inspection and quality control in production (defects, holes, production lanes)
  • and so on ...

Here are some examples:

Data Statistics

• All images have the same resolution 4032 x 2268 x 3 (height x width x channels)

• There are no labeled objects

• Datasets:

  • 01_background - background images without objects
  • 02_sunflower - sunflower seeds with different density
  • 03_pumpkin_peeled - peeled pumpkin seeds with different density
  • 04_pumpkin_unpeeled - unpeeled pumpkin seeds with different density
  • 05_pumpkin_mx - mix of peeled and unpeeled pumpkin seeds with different density
  • 06_mix - mix of sunflower, peeled and unpeeled pumpkin seeds

• Classes