I have an unlabeled dataset: 500 X-rays of seeds in husk. In each image there are different number of seeds, for instance, from $10$ to $50$.

enter image description here

A seed has some features. The main features are a seeds' square and a husks' square in pixels. I need to calculate the these features ratio.

I am looking for an objects recognize algorithm. My problem is: I don't know what is a task type.

enter image description here

The dataset is unlabeled and I can use unsupervised learning and identify a similarity in groups (clustering task). Or I should to label data and use supervised learning and identify a real values (regression task)?

Question What is a task type?


  1. The Plant Pathology 2020 challenge dataset to classify foliar disease of apples
  2. Exploring Survival on the Titanic
  • $\begingroup$ @D.W., I have updated the question. $\endgroup$ – Nick Nov 12 '20 at 0:22
  • $\begingroup$ Thank you for the edits. What do you mean by "an objects recognition quality"? Can you edit the question to define or explain what you mean by that phrase? Our usual rule is to ask one question per post - if you have multiple questions, it is often better to ask them separately. $\endgroup$ – D.W. Nov 12 '20 at 6:19
  • $\begingroup$ Regarding algorithms, have you tried anything? Have you done any research? What approaches have you considered? The site works best when you do research before asking and show us what you've found so far and why you've rejected it. Object detection in particular can require some trial-and-error so it's helpful to review standard approaches, try them on your images, and describe how well they work and any shortcomings that you want to focus on. $\endgroup$ – D.W. Nov 12 '20 at 6:20
  • $\begingroup$ @D.W. I reformulate the question. $\endgroup$ – Nick Nov 15 '20 at 3:56
  • 1
    $\begingroup$ Where did you encounter the phrase "task type"? Perhaps you should read a different resource about machine learning, if it uses a phrase you can't understand. $\endgroup$ – D.W. Nov 15 '20 at 6:55

This is likely to be a challenging task, and if you're not familiar with image classification, I suspect it may be beyond what you can reasonably do right now.

I would suggest you investigate classical image processing techniques. You might try applying morphological operators, such as closing (dilation followed by erosion) or opening followed by thresholding. You may need to experiment. If you search on Stack Overflow and Signal Processing sites, you'll find many examples.

See also blob detection and watershed segmentation.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.