Deep learning based banana plant detection and counting using high-resolution red-green-blue (RGB) images collected from unmanned aerial vehicle (UAV)

Abstract

The production of banana—one of the highly consumed fruits—is highly affected due to loss of certain number of banana plants in an early phase of vegetation. This affects the ability of farmers to forecast and estimate the production of banana. In this paper, we propose a deep learning (DL) based method to precisely detect and count banana plants on a farm exclusive of other plants, using high resolution RGB aerial images collected from Unmanned Aerial Vehicle (UAV). An attempt to detect the plants on the normal RGB images resulted less than 78.8% recall for our sample images of a commercial banana farm in Thailand. To improve this result, we use three image processing methods—Linear Contrast Stretch, Synthetic Color Transform and Triangular Greenness Index—to enhance the vegetative properties of orthomosaic, generating multiple variants of orthomosaic. Then we separately train a parameter-optimized Convolutional Neural Network (CNN) on manually interpreted banana plant samples seen on each image variants, to produce multiple results of detection on our region of interest. 96.4%, 85.1% and 75.8% of plants were correctly detected on three of our dataset collected from multiple altitude of 40, 50 and 60 meters, of same farm. Further discussion on results obtained from combination of multiple altitude variants are also discussed later in the research, in an attempt to find better altitude combination for data collection from UAV for the detection of banana plants. The results showed that merging the detection results of 40 and 50 meter dataset could detect the plants missed by each other, increasing recall upto 99%.

A Cloud-Based Environment for Generating Yield Estimation Maps From Apple Orchards Using UAV Imagery and a Deep Learning Technique

“…On the other hand, with visual counting (Nfp), considered to be the most reliable method, an accuracy of 88.96% was obtained. The errors between visual counts and object detection were similar to those obtained by Neupane et al (2019) when counting banana fruits. These results demonstrate that the use of simple data augmentation techniques such as picture rotation, filters, and transfer learning can facilitate the building of tools with a high potential for apple fruit detection.…”

Fast and Accurate Detection of Banana Fruits in Complex Background Orchards

“…The detection of pomelo [44], kiwi fruit [45], waxberry [46], guava [47], and other fruits have been gradually concerned; With the development of deep learning, fruit flower detection, which is difficult to the traditional algorithm, has been emerging [48][49][50][51]. In the detection of vegetables, the improvement in the bounding box and the detection rate is the research focus of the tomato detection network [52][53][54]; Based on deep neural network, excellent results have been achieved in cucumber fruit length estimation [55], sweet pepper detection [56], date fruit variety and maturity judgment [57] and other aspects. In recent years, deep convolutional neural network has been applied in the banana plantation.…”

“…In recent years, deep convolutional neural network has been applied in the banana plantation. Based on fast-RCNN, NeupaneI et al [57] recognized and counted banana plants on the farm by using RGB aerial images collected by UAV. Clark et al [58] detected banana plantations through aerial images and used U-NET neural network to draw maps, but did not conduct detection and research on banana fruits.…”

Oil Palm Detection via Deep Transfer Learning

“…The rise of Artificial Intelligence (AI) models for image analysis such as deep learning and associated methods has attracted the attention of researchers in the field of precision agriculture [10] with a particular emphasis on embedding it in the oil palm sector where an informed use of AI in the oil production process can lead to major improvements with high economical, environmental and suitability impact [8], [11], [12]. Indeed, several studies are analysing spectral images with AI tools for the identification of crop units [13], [14].…”