case
Collection and segmentation of plants from drone aerial imagery of plantations for monitoring the condition of crops

Agriculture Data labeling

The use of machine learning technologies to automate and optimise the processes in agricultural enterprises
Agriculture
Dividing an image into semantically independent segments to determine object boundaries
Segmentation
Process of recognition and grouping of objects into preset categories
Classification
The process of identifying objects to train systems to recognize and interpret them
Data Labeling
Case description
Data Collection:
Plantations of root crops and vegetables (cabbage, zucchini, potatoes, turnips, etc.) were collected using drones from a bird's-eye view.
Data Labelling:
1. Semantic segmentation for determining the class of cultures

2. Instance segmentation by polygons to determine the different state of crops in the image

3. The keypoints labelling is represented by coordinates in .xml file
Depending on the complexity of annotation and the required quality, we place the data in CVAT or PhotoShop, apply AI pre-labelling and conduct quality control before transmitting the data.
Tools:
AI solutions for the agricultural sector:
1.
Mapping of fields and plantations with indication of favorable areas and conditions
2.
Control of the automatic irrigation system to prevent the death of crops, as well as setting up automatic spraying of fertilizers and pesticides
3.
Search and detection of weeds between beds in large spaces
4.
Analysis and assessment of the state of plantings
5.
Monitoring of agricultural land to search for plants affected by parasites, injured or killed due to drought or excessive watering
6.
Tracking the growth and development of crops to determine the degree of maturity of fruits
Didn't find the necessary information?
Leave a request to calculate the cost of a dataset for your business!
Tell us about your project!
Our client's project:
Laboro Tomato - is an image dataset of growing tomatoes at different stages of their ripening which is designed for object detection and instance segmentation tasks. We also provide two subsets of tomatoes separated by size