Advancing Bag-of-visual-words Representations For Lesion Classification In Retinal Images.

Diabetic Retinopathy (DR) is a complication of diabetes that can lead to blindness if not readily discovered. Automated screening algorithms have the potential to improve identification of patients who need further medical attention. However, the identification of lesions must be accurate to be useful for clinical application. The bag-of-visual-words (BoVW) algorithm employs a maximum-margin classifier in a flexible framework that is able to detect the most common DR-related lesions such as microaneurysms, cotton-wool spots and hard exudates. BoVW allows to bypass the need for pre- and post-processing of the retinographic images, as well as the need of specific ad hoc techniques for identification of each type of lesion. An extensive evaluation of the BoVW model, using three large retinograph datasets (DR1, DR2 and Messidor) with different resolution and collected by different healthcare personnel, was performed. The results demonstrate that the BoVW classification approach can identify different lesions within an image without having to utilize different algorithms for each lesion reducing processing time and providing a more flexible diagnostic system. Our BoVW scheme is based on sparse low-level feature detection with a Speeded-Up Robust Features (SURF) local descriptor, and mid-level features based on semi-soft coding with max pooling. The best BoVW representation for retinal image classification was an area under the receiver operating characteristic curve (AUC-ROC) of 97.8% (exudates) and 93.5% (red lesions), applying a cross-dataset validation protocol. To assess the accuracy for detecting cases that require referral within one year, the sparse extraction technique associated with semi-soft coding and max pooling obtained an AUC of 94.2 ± 2.0%, outperforming current methods. Those results indicate that, for retinal image classification tasks in clinical practice, BoVW is equal and, in some instances, surpasses results obtained using dense detection (widely believed to be the best choice in many vision problems) for the low-level descriptors.

Sistema de monitoramento de perdas visíveis de cana-de-açúcar em colhedora de cana picada

In the last few years the sugar-cane mechanical harvested area has increased, especially in regions with appropriated slop. The use of this technology brings some inconveniences, such as, the increase in the percentage of extraneous matter, which causes the reduction of technological quality of the raw material, and losses in the field. Extraneous matter (trash) is composed of tops and leaves in major percentage, plus soil and roots, and eventually some metal parts. In the green cane harvest system the percentage of extraneous matter has a tendency to increase due to the great amount of vegetal matter to be processed. The increase in the blower fan speed to reduce the amount of extraneous matter can lead to an unacceptable economic level of raw material losses. The main objective of this work was, using a cane loss monitor, to evaluate and quantify the amount of visible losses of sugar cane through the primary extractor at two different fan speeds. Afterwards these losses were related to the harvester cleaning efficiency. The piezoelectric transducer shows a reasonable sensibility. The results show that the cleaning efficiency in the primary extractor (85% mean), the cane losses (between 5.68% and 2.15%) and fan speed are interrelated. The total losses and specially splinters (between 3.19% and 0.91%), showed a significant difference among the treatments.

Simulação da varredura da superfície por um cortador de base segmentado

Base cutting and feeding into harvesters of plants lying close to the ground surface require an efficient sweeping action of the cutting mechanism. It is not the case of conventional sugarcane harvesters which have rigid blades mounted on discs capable to contaminate the cane with dirt as well as damage the ratoons. The objective of this work was to simulate the sweeping performance of a segmented base cutter. The model was developed using the laws of dynamic. Simulation included two rotational speeds (400 and 600 rpm), two cutting heights (0.12 and 0.13 m) and two disk tilting angles (-10º and -12º). The simulated sweeping angle varied between 56º and 193º, which are very promising as a mean to cutting and feeding cane sticks lying on the ground. Cutting height was the variable that affected sweeping action the most. This behavior indicates the need to have an automatic control of the cutting disk height in order to keep good sweeping performance as the harvester moves forward.

Influência do auxílio mecânico na colheita de tomates

Fresh tomato harvest is traditionally made without harvesting aids. The main goal of this research was to evaluate performance parameters of fresh tomato harvesting aid equipment and compare it to traditional harvest, in the state of São Paulo. Therefore, an equipment was developed and the harvest process was evaluated in four different ways: traditional system (harvest system used in Santa Luzia farm, Brotas, SP, Brazil), picker walking with a harvesting aid equipment, picker seated in a harvesting aid equipment and a composition of both systems: two pickers seated and one picker walking in two different velocities ranges. The different systems using harvesting aid showed an average yield by picker more efficient than reference. Harvest system using three pickers showed an increase of 290% on yield average by picker, on the range of 0.5-1.0 fruit per plant, followed by the systems with a walking picker, that increased productivity in 41%, and picker seated harvester, that showed an increase of 35%. These results demonstrate the importance of using a harvesting aid equipment.

Avaliação de perdas invisíveis na colheita mecanizada em dois fluxos de massa de cana-de-açúcar

In this work the performance of a sugar cane chopped harvester was analysed when fed with two sugar cane mass flows, measuring the invisible losses, which are impossible to measure in the field, harvester sugar cane cleaning efficiency and air velocity on extractors exit. The trial was done under controlled conditions at Copersucar Technology Center in January 2000. The results showed that the flow of sugar cane through the harvester doesn't influence the magnitudes of total invisible losses and raw material cleaning efficiency. The mean air velocity on the primary extractors exit was 12.0 m s-1, and 9.2 m s-1 on the secondary extractor, with a coefficient of variation of 21%, indicating that the poor cleaning performance of the harvester could be related to air velocity difference inside the extractor. Analyzing the data collected in the trials, it was possible to conclude that invisible losses in sugar cane harvester were 10% and the cleaning efficiency was 87%.

Administração limpa e enxuta em sistemas hidráulicos de colhedoras de cana-de-açúcar

Among the important changes in the production processes, it is necessary to guarantee the sustainability of the human enterprises, what makes us to foresee changes in the managerial administration to adapt to a new model, with the insert of the concepts of Clean Production, Cleaner, Lean and Total Productive Maintenance (TPM). The main focus of this work was to elaborate a methodology that made it possible to guarantee the reliability in the waterworks of the sugarcane harvester, identifying and analyzing the manners of flaws, in order to result in the improvement of the environmental and socioeconomic quality in the atmosphere of an industry of sugarcane through the significant decrease of hydraulic oil spill. Through the existent report in ERP (Enterprise Resource Planning), used in a Sugarcane Industry Plant, it was possible to accompany of the operational acting of the sugarcane harvester used during 03 crops, regarding the manners of flaws in the waterworks of the same ones, and, in one of the crops it was elaborated the total control of the waterworks of 5 harvesters. Based on the obtained data and the developed methodology it was possible to develop a software that specifies the electric outlet of decisions.