rezacsedu / Rice-crop-Insects-and-Weed-Detection-using-faster-R-CNN

As the increase in the world population the demand of the rice is also increases. In order to increase the growth of rice in the rice crop it is necessary to detect the weed and insects in the rice crop to minimize the growth of weed and insects so that the growth of the rice can be increased.Insect and Weed detection is the important factor to be analyzed. Unmanned Air Vehicle (UAV) is used for data acquisition of rice crop in different phases and states so that high quality of RGB images can be captured. In which we have taken 15 different types of rice crop insects species images and different phases of weed images to train the model. The proposed method facilitates the extraction of weed and insects into the rice crop field using deep learning concept faster region-based convolutional neural networks(Faster R-CNNs) it is implemented using Python3 with the help of Tensorflow API. The result shows that Faster R-CNN method is the state of arts method for detection and classification of weed and insects with good accuracy rate.

INSECTS AND WEED DETECTION AND CLASSIFICATION FROM RICE CROP FIELD USING FASTER R-CNN

Abstract

As the increase in the world population the demand of the rice is also increases. In order to increase the growth of rice in the rice crop it is necessary to detect the weed and insects in the rice crop to minimize the growth of weed and insects so that the growth of the rice can be increased.Insect and Weed detection is the important factor to be analyzed. Unmanned Air Vehicle (UAV) is used for data acquisition of rice crop in different phases and states so that high quality of RGB images can be captured. In which we have taken 15 different types of rice crop insects species images and different phases of weed images to train the model. The proposed method facilitates the extraction of weed and insects into the rice crop field using deep learning concept faster region-based convolutional neural networks(Faster R-CNNs) it is implemented using Python3 with the help of Tensorflow API. The result shows that Faster R-CNN method is the state of arts method for detection and classification of weed and insects with good accuracy rate.

Keywords :-

Computer vision, Deep learning, Weed detection, Insects detection, Rice detection, Convolutional neural networks, Region based convolutional neural networks

Problem Statement

Objects contained in image files can be located and identified automatically. This is called object detection and is one of the basic problems of computer vision. As we will demonstrate, region-base convolutional neural networks(Faster R-CNN) are currently the state-of-theart solution for object detection. The main task of this thesis is to detection and classification of rice crop insects and weed into rice crop field with the help of images and videos which is taken by UAV

User Manual

In order to excess project people need to go inside models directory in which we will find some short of directory from them we need to find research directory and then netx,we need to find object_detection directory and after that all these hierarchical directories we need to follow the instruction given on that page README.md. Again, i am showing steps here which we need to follow them in order to use/excess this project source code.

Step 1 : Open models directory

Step 2 : Open research directory

Step 3 : Open object_detection directory

Step 4 : Follow the steps given on that page README.md

Thank a lot and Regards...

Radhe Raman Tiwari