Using computer vision to categorize tyres and estimate the number of visible tyres in tyre stockpile images
- Authors: Eastwood, Grant
- Date: 2017
- Subjects: Tires -- Specifications Tires -- Recycling , Tires -- Maintenance and repair
- Language: English
- Type: Thesis , Masters , MSc
- Identifier: http://hdl.handle.net/10948/16022 , vital:28313
- Description: Pressures from environmental agencies contribute to the challenges associated with the disposal of waste tyres, particularly in South Africa. Recycling of waste tyres in South Africa is in its infancy resulting in the historically undocumented and uncontrolled existence of waste tyre stockpiles across the country. The remote and distant locations of such stockpiles typically complicate the logistics associated with the collection, transport and storage of waste tyres prior to entering the recycling process. In order to optimize the logistics associated with the collection of waste tyres from stockpiles, useful information about such stockpiles would include estimates of the types of tyres as well as the quantity of specific tyre types found in particular stockpiles. This research proposes the use of computer vision for categorizing individual tyres and estimating the number of visible tyres in tyre stockpile images to support the logistics in tyre recycling efforts. The study begins with a broad review of image processing and computer vision algorithms for categorization and counting objects in images. The bag of visual words (BoVW) model for categorization is tested on two small data sets of tread tyre images using a random sub-sampling holdout method. The categorization results are evaluated using performance metrics for multiclass classifiers, namely the average accuracy, precision, and recall. The results indicated that corner-based local feature detectors combined with speeded up robust features (SURF) descriptors in a BoVW model provide moderately accurate categorization of tyres based on tread images. Two feature extraction methods for extracting features for use in training neural networks (NNs) for tyre count estimations in tyre stockpiles are proposed. The two feature extraction methods are used to describe images in terms of feature vectors that can be used as input for NNs. The first feature extraction method uses the BoVW model with histograms of oriented gradients (HOG) features collected from overlapping sub-images to create a visual vocabulary and describe the images in terms of their visual word occurrence histogram. The second feature extraction method uses the image gradient magnitude, gradient orientation, and edge orientations of edges detected using the Canny edge detector. A concatenated histogram is constructed from individual histograms of gradient orientations and gradient magnitude. The histograms are then used to train NNs using backpropogation to approximate functions from the feature vectors describing the images to scalar count estimations. The accuracy of visible object count predictions are evaluated using NN evaluation techniques to determine the accuracy of predictions and the generalization ability of the fit model. The count estimation experiments using the two feature extraction methods for input to NNs showed that fairly accurate count estimations can be obtained and that the fit model could generalize fairly well to unseen images.
- Full Text:
- Date Issued: 2017
Yield responses, mineral levels of forages and soil in old arable land planted to four legume pasture species in Lushington communal area, South Africa
- Authors: Gulwa, Unathi
- Date: 2017
- Subjects: Forage plants -- South Africa -- Eastern Cape Minerals in animal nutrition Communal rangelands -- South Africa -- Eastern Cape
- Language: English
- Type: Thesis , Masters , MSc
- Identifier: http://hdl.handle.net/10353/2799 , vital:28091
- Description: This study was conducted in the old arable land located in Lushington communal area in the Eastern Cape province of South Africa. The objectives of the study were to assess the effect of legume introduction on biomass yield, forage and soil mineral levels of the arable lands planted to four leguminous pastures in four seasons. Planting was done in March and October 2008 in Lushington. All legumes were subjected to grow under rain fed conditions. Trifolium vesiculosum (arrowleaf clover), Lespedeza cuneata (sericea lespedeza), Trifolium repens (white clover) and Lotus corniculatus (birdsfoot trefoil) are the four forage legume species that were sampled for the purposes of this study. The four legume species persisted out of the fourteen species that were initially tested for adaptability and persistence in the environmental conditions of Lushington communal area. The legumes, grasses and soils from these legume plots were sampled to determine the effect of legume introduction on the forage yield, mineral contents of the companion grasses and soils over four seasons. Plant and soil samples were collected once in spring (November) 2013, summer (February), autumn (March) and winter (May) 2014 for biomass production, macro and micronutrients determination. Results indicated that legume inclusion and season affected (P < 0.05) the total dry matter (TDM) yield production. Plots with Lespedeza cuneata had the highest TDM (1843 kg/ha) and control plots had the least dry matter production (1091 kg/ha). Summer season provided the highest (P < 0.05) TDM compared to the other seasons. Both legume and grass quality was also affected (P < 0.05) by legume inclusion in different seasons. Accordingly, grasses harvested from Trifolium repens plot showed higher CP level (10.90 percent) than those harvested from other plots whereas the lowest grass CP content (6.90 percent) was measured in the control treatment. L. cuneate had the highest (P < 0.05) CP level (11.00 percent) and T. repens had the least CP (6.63 percent) level. Grasses harvested in autumn had the highest (P < 0.05) CP level (12.50 percent) and those harvested in winter had the least CP level (4.60 percent). Similarly, all legume pastures harvested in spring had superior (P < 0.05) CP (10.80 percent) levels and those harvested in winter had the least CP (3.50 percent) level. Legume inclusion had an effect (P < 0.05) on both grass and legume macro nutrient contents. Trifolium repens plot had the highest grass K (1.07 percent), Ca (1.50 percent) and Mg (1.83 percent), whereas there were lower K (0.12 percent), Ca (1.25 percent) and Mg (1.08 percent) contents in grasses harvested from the control and T. vesiculosum plots, respectively. In legumes, macro nutrient concentrations: K (0.68 percent), Ca (1.75 percent) were superior in the T. vesiculosum plot in comparison to other plots. Season also affected (P < 0.05) both grass and legume macro nutrient content. There was higher K (0.90 percent), Ca (1.30 percent) and Mg (0.94 percent) content in grasses harvested in autumn whereas there were lower levels in winter harvests. In legumes, superior K (0.74 percent) and Mg (1.87 percent) content were attained during spring while the least were measured in winter (0.07 percent) and autumn (0.75 percent), respectively. Likewise, both legume inclusion and season had an significant effect (P < 0.05) on the forages micronutrient levels. During spring, there was superior soil P content (36.28 mg/kg) while during autumn; there was less P (22.58 mg/kg) content. The highest SOC level (1.49 percent) was measured in the T. repens plot whereas the lowest SOC (1.15 percent) was attained in the control plot. The results of this study showed that grass legume mixtures produced forages with high nutrient content and herbage yield. Legume planting in the old arable lands has a potential to improve soil quality parameters such as soil P and SOC content.
- Full Text:
- Date Issued: 2017