Canopy Coverage Estimation & Tassel Detection Using UAV Based Remote Sensing and Crop-Weed Segmentation Using Static Images

Taparia, Mahesh and Rajalakshmi, P (2019) Canopy Coverage Estimation & Tassel Detection Using UAV Based Remote Sensing and Crop-Weed Segmentation Using Static Images. Masters thesis, Indian institute of technology Hyderabad.

[img] Text
Restricted to Repository staff only until 26 June 2024.

Download (6MB) | Request a copy


By 2050, agriculture production needs to double in order to reach the food demand due to rise in population. Breed scientists are focusing on the breed of crop which can improve the nutrient quality and yield of a crop. There is a need to bring innovation in agriculture management in order to fulfill the food demand and to secure the food for the next generation with the limitation of natural resources like land, fresh water, etc. Precision farming is the method which uses different technologies in order to meet the requirement of good health and productivity of crop with optimal use of agronomic inputs like water, pesticides, and fertilizers. This thesis addressed some of the challenges of agriculture like a selection of best breed of the crop, optimal use of agronomic inputs, crop health management, crop yield, etc. and provides the solution based on image processing and deep learning for precision agriculture. In order to manage the big field properly, currently the farmers used to visit the whole field and give the proper treatment. This manual work is a time consuming and labour intensive task. So to reduce this tedious task, drone based remote sensing is going to bring a revolution in agriculture industry. Along with the different emerging technical fields like machine learning, deep learning, computer vision, etc., drone can be a key factor in accelerating the study of different phenotyping and genotyping traits of a crop. Drone mounted with different sensors like RGB camera, multi-spectral camera, thermal camera, lidar etc. is helping in analysis of different physical traits of the crops. Multi-spectral camera have NIR (near infra-red) as well as visible spectrum. Periodic survey of fields with these sensors is helps in evaluation of crop growth and production. Periodically, we have taken the images from the field and we have targeted the rice and maize crop for our analysis. We have analyzed the 216 different breeds of rice and estimated the optimum variety with a good growth with the help of drone images. Also, the problem of presence of weed in the field is analyzed using static images and a deep learning based framework is used to discriminate crop and weed. The different growth stage of maize is also studied and we have estimated the heading stage and days to 50% flowering using drone based remote sensing.

[error in script]
IITH Creators:
IITH CreatorsORCiD
Item Type: Thesis (Masters)
Uncontrolled Keywords: UAV, Computer Vision, Deep Learning, Phenotyping
Subjects: Electrical Engineering
Divisions: Department of Electrical Engineering
Depositing User: Team Library
Date Deposited: 26 Jun 2019 11:24
Last Modified: 26 Jun 2019 11:24
Publisher URL:
Related URLs:

    Actions (login required)

    View Item View Item
    Statistics for RAIITH ePrint 5564 Statistics for this ePrint Item