DEEP LEARING BASED WEED DETECTION IN DIGITAL IMAGES

Abstract

In plantation we are recognizing the uneven plant spacing between saplings of the crop and they are growing in
the same way. Due to uneven plant spacing the researchers the facing many difficulties to identify the weeds int the
vegetable plantation. We can find different species of weeds in our crop/plant such as Black Grass, Fat hen, Clevers,
Charlock and many more. The aim of this research is to identify these kinds of weeds using the Image processing and deep
learning techniques. In deep learning we have used the Alex Net model for the purpose of identifying the weeds and in
addition Alex Net model we have added image processing techniques which helps the model in performing the operations
effortlessly. We are using Python Version 3.11 to identify the weeds.
Firstly, we are training the Alex Net model to identify some sort of weeds and then we are fetching the input image to
image preprocessor which simplifies the image data present in the input image by enhancing some image features or by
suppressing the unwilling distortions important for further processing, the simplified image is carried on to the next block of
preprocessing which is the Feature Extraction . Then the output the feature extraction block is fetched to the pre-trained Alex
Net model as input where it classifies the given data and then image segmentation is performed by which it converts an
image into a collection of regions of pixels and these are represented by a mask. After image segmentation the data is sent to
the identification block where the species of weed are identified. The pre-trained