RESUMO
Many researchers have realized the intelligent medical diagnosis of diabetic retinopathy (DR) from fundus images by using deep learning methods, including supervised contrastive learning (SupCon). However, although SupCon brings label information into the calculation of contrastive learning, it does not distinguish between augmented positives and same-label positives. As a result, we propose the concept of Angular Margin and incorporate it into SupCon to address this issue. To demonstrate the effectiveness of our strategy, we tested it on two datasets for the detection and grading of DR. To align with previous work, Accuracy, Precision, Recall, F1, and AUC were selected as evaluation metrics. Moreover, we also chose alignment and uniformity to verify the effect of representation learning and UMAP (Uniform Manifold Approximation and Projection) to visualize fundus image embeddings. In summary, DR detection achieved state-of-the-art results across all metrics, with Accuracy = 98.91, Precision = 98.93, Recall = 98.90, F1 = 98.91, and AUC = 99.80. The grading also attained state-of-the-art results in terms of Accuracy and AUC, which were 85.61 and 93.97, respectively. The experimental results demonstrate that Angular Margin is an excellent intelligent medical diagnostic algorithm, performing well in both DR detection and grading tasks.
RESUMO
We design and propose a compact street lamp based on dual-module chip-on-board LED. The street lamp is composed of six faceted reflectors. It can direct the luminous flux and form uniform illumination on the target area, and it effectively reduces power consumption. We have conducted both simulations and prototype measurements. The test results show good optical performance in that the uniformity of luminance reaches 0.58 for LED lamp zigzag arrangements and 0.60 for LED lamp double-side arrangements. The average luminance can fulfill the requirements in Chinese road lighting Standard CJJ45-2006.
RESUMO
A hybrid illumination method using both daylight and light-emitting diodes (LEDs) for indoor lighting is presented in this study. The daylight can be introduced into the indoor space by a panel-integration system. The daylight part and LEDs are combined within a specific luminaire that can provide uniform illumination. The LEDs can be turned on and dimmed through closed-loop control when the daylight illuminance is inadequate. We simulated the illumination and calculated the indoor lighting efficiency of our hybrid daylight and LED lighting system, and compared this with that of LED and fluorescent lighting systems. Simulation results show that the efficiency of the hybrid daylight/LED illumination method is better than that of LED and traditional lighting systems, under the same lighting conditions and lighting time; the method has hybrid lighting average energy savings of T5 66.28%, and that of the LEDs is 41.62%.
RESUMO
This paper proposes an LED-based compound optical system, which can be involved in the design of the driving beam system in automotive headlamps with high system efficiency and low power consumption. The compound system can meet the requirements announced in the UNECE regulation "Addendum 111: Regulation No. 112 Revision 2." Also, it is composed of a nonspherical reflector, a compound lens, and a two-dimensional diverging lens. Using a single device of high-brightness LED of merely 7.6 W, the specified illumination requirements for the driving beam can be achieved. As we have expected, on the test screen at a distance of 25 m, the simulation results, as well as the testing results for the prototype, can reach the illuminance distribution requirements, including all specified regions and key points. Moreover, this compound system enjoys the features of high compactness, high energy efficiency, and feasibility of manufacturing.
RESUMO
This paper proposes a low-beam system for an LED-based headlamp architecture, which is composed of an elliptical reflector, a baffle, and a faceted reflector. Using a single device with high brightness LED of merely 6.00 W, two devices total 12.00 W. With a low beam 55 W traditional halogen light source compared to 78.18% energy savings, the specified illumination requirements for the headlamp low beam can be achieved, according to the ECE regulation "Addendum 111: Regulation No. 112 Revision 2." As we have expected, on the test screen at a distance of 25 m, the simulation results as well as the testing results for the prototype can reach the illuminance distribution requirements, including all specified regions and key points. Moreover, this faceted low beam system enjoys the features of high compactness, high energy efficiency, and feasibility of manufacturing.
RESUMO
This Letter proposes an idea of a natural light illumination system by using a light collecting panel. The light collecting panel is coupled with a light guide followed by a specifically designed lens to achieve uniform illumination on a target screen. To assess the system specification, we build a primary model in three-dimensional modeling software and make ray tracing simulation based on a Monte Carlo algorithm. The simulation results show a relatively high efficiency of 41.41% for the entire system and achieve a high uniformity of more than 0.8 within the circular illumination area.
RESUMO
The precision approach path indicator (PAPI) system is installed alongside a runway or at a heliport to provide accurate visual approach path information to pilots of landing aircraft, either day or night. The indicator sends the information through different color combinations. The international standards have very rigid requirements for the intensity distribution and the transition angle from red light to white light of the indicator. These two parameters are the main parameters to judge the accuracy of the indicator. We propose a new method to measure these two optical parameters, which can improve the accuracy of the signals and guarantee safe landing of the aircraft.
