Articles, Conference and Workshop Papers Collection
Permanent URI for this collectionhttp://10.10.97.169:4000/handle/123456789/4223
Browse
Browsing Articles, Conference and Workshop Papers Collection by Author "Fue, Kadeghe G."
Now showing 1 - 2 of 2
- Results Per Page
- Sort Options
Item Deep learning based Real-time GPU-accelerated tracking and counting of cotton bolls under field conditions using a moving camera(2018 ASABE Annual International Meeting, 2018-08) Fue, Kadeghe G.; Porter, Wesley; Rains, GlenRobotic harvesting involves navigation and environmental perception as first operations before harvesting of the bolls can commence. Navigation is the distance required for a harvester’s arm to reach the cotton boll while perception is the position of the boll relative to surrounding environment. These two operations give a 3D position of the cotton boll for picking and can only be achieved by detection and tracking of the cotton bolls in real-time. It means detection, tracking and counting of cotton bolls using a moving camera allows the robotic machine to harvest easily. GPU-accelerated deep neural networks were used to train the convolution networks for detection of cotton bolls. It was achieved by using pretrained tiny yolo weights and DarkFlow, a framework which translates YOLOv2 darknet neural networks to TensorFlow. A method to connect tracklets using vectors that are predicted using Lucas-Kanade algorithm and optimized using robust L-estimators and homography transformation is proposed. The system was tested in defoliated cotton plants during the spring of 2018. Using three video treatments, the counting performance accuracy was around 93% with standard deviation 6%. The system average processing speed was 21 fps in desktop computer and 3.9 fps in embedded system. Detection of the system achieved an accuracy and sensitivity of 93% while precision was 99.9% and F1 score was 1. The Tukey’s test showed that the system accuracy and sensitivity was the same when the plants were rearranged. This performance is crucial for real-time robot decisions that also measure yield while harvesting.Item A simple convolutional neural network architecture for monitoring Tuta absoluta (Gelechiidae) infestation in tomato plants.(Sokoine University of Agriculture, 2021-05) Mourice, Sixbert K.; Mlebus, Festo Joseph; Fue, Kadeghe G.Tomato leaf miner (Tuta absoluta (Gelechiidae)) is a serious tomato insect pest in Tanzania, and its management or control still posess significant challenge. If left uncontrolled, the loss inflicted by the miner can be as high as 100%. Successful management of the pest may leverage on an integrated pest management (IPM) approach which, requires high throughput data on damage signs over space and time. This needs, in turn, a robust technique for pest monitoring. This study uses a deep learning technique to detect infestation symptoms of T. absoluta on tomato plants. The technique is rapid, automated and doesn’t require trained or experienced personnel. An experiment was carried out at Sokoine University of Agriculture (SUA), where two sets of tomato plants (cv. Asila F1) were planted in a screen house and in an open field. High-quality images of the tomato leaves were captured from both sets at seven days intervals for 70 days following transplanting. More images were collected from tomato gardens around Morogoro town. Collected images were labeled as being infested or non-infested. A simple convolution neural network (CNN) architecture with four convolution layers, three pooling layers, one flat layer and one dense layer, powered by Keras library and python’s Tensorflow backend, was developed in R-Software. The model accuracy was 90% on training and 82% on test data sets. This study suggests that the model can accurately identify T. absoluta infestation in tomato plants to a considerable extent. An in-depth discussion of the technique is provided in the paper.