Find similar products
Run the project in the Sandbox first. This helps you avoid making mistakes and spending money on a task that isn't working right.
This is an example of taking a task for collecting relevant data and decomposing it into multiple projects. Decomposition can improve the quality of results and reduce the cost of performing complex tasks.
Collecting relevant data is a task to search for matching data. Let's say you have photos of clothes and shoes and you need to find similar shoes in an online store. But you don't know if there are shoes in all the photos and want to filter them first. Next, you need to find similar shoes in the online store. You also want to check the performers' responses yourself afterwards. To get the final result, you need to compare the found shoes with the original image and decide which pair is most similar.
You can use the projects described below either together to set up a unified process or just in part (for example, to only identify the object in the photo).
- Project 1: Use this project if not all images contain the object you are looking for and you want to filter them.
- Project 2: In this project, performers will search in an online store for shoes that are similar to the ones shown. Use the results from project 1 as source images.
- Project 3: In this project, performers will compare the shoes from the original image (in project 1) with the shoes found in project 2.
- Project 4: In this project, performers will compare the found shoes with the original image and decide which pair is most similar.
See the complete code for all projects in the Appendix.
Upload a TSV file with tasks to the pool.
Start the pool.
Obtain and aggregate the results.