Increasing the Effectiveness of the Weeding and Harvesting Process

Services provided: Image Annotation

Published date: 22.05.2023

Read time: 3 min

Client Profile

Industry: Agriculture
Location: USA
Size: 300+

Company Bio

The client is a manufacturer of robotic equipment for the agricultural industry. Their products are geared towards regenerative farming without additional labor, which reduces costs to farmers and eases contained resources. They are working to reduce and eliminate synthetic inputs to our agricultural systems

Services Provided

Project Overview

Mindy Support helped the client increase the accuracy of their weeding and harvesting robot with high-quality data annotation.

Business Problem

The client was looking to increase the quality of their weeding and harvesting robot and prepared a sizable dataset of images to be used as training data for the ML algorithms. The images contained rows within a field that the robot would need to travel through, and this path needed to be annotated with a lot of precision. For this reason, the client understood that the chosen service provider would need to have proven processes in place to make sure the needed quality level was reached.

Why Mindy Support

The client announced a tender for the contract, and Mindy Support was selected to create a pilot project with a test batch of data. The client was very impressed with the quality delivered and decided to expand the project to include additional data.

Solutions Delivered to the Client

Our team understood the importance of quality to the overall success of the project, so we set up rigorous processes to make sure that the necessary quality rate was achieved. The pace of all the annotation work needed to be done was lightning-quick, but the quality of the annotations needed to remain high. In order to reach the quality level demanded by the client, we designed and implemented a completely new workflow, which had 4 steps:

  1. The annotator marked a row
  2. The approver’s team made an initial check
  3. The QA team made a second check
  4. The Quality manager made a final check of the labels

At every stage, Mindy Support’s team members made corrections, as needed, which produced a very accurate and precise result. While it may seem that the workflow mentioned above was extensive or overly convoluted, we managed to set everything up in a way that everything was done by the client’s deadline. When the client saw the outstanding results of our work, they found the workflow quite reasonable.

Key Results

  • 2 years of overall experience with the project
  • turnaround time – 24 hours
  • 10+ acres of fields annotated
  • 25 agents on the project

TABLE OF CONTENTS

    Stay connected with our latest updates by subscribing to our newsletter.

      ✔︎ Well done! You're on the list now

      GET A QUOTE FOR YOUR DATA LABELING PROJECT

        I have read and agree to the Privacy Policy

        We have a minimum threshold for starting any new project, which is 735 productive man-hours a month (equivalent to 5 graphic annotators working on the task monthly).

        TALK TO OUR EXPERTS ABOUT YOUR AI/ML PROJECT

        CONTACT US