Yazar "Isikdag, U." seçeneğine göre listele

Listeleniyor 1 - 3 / 3
  • [ X ]
    Öğe
    CLASSIFICATION OF HAZE IN CITY IMAGES WITH CONVOLUTIONAL NEURAL NETWORKS AND TRANSFER LEARNING
    (Scibulcom Ltd, 2021) Isikdag, U.; Apak, Sudi
    Air quality has an enormous impact on health. To take preventive measures on time, it is important to track and estimate air pollution. In the estimation of air pollution, the data acquisition from images is easy and of low-cost, when compared with sensor-based data acquisition. Machine and Deep Learning methods utilise images and videos from city cameras or social media and provide accurate estimations of air pollution. In this context, the aim of this study was testing the accuracy and efficiency of Deep Learning and Convolutional Neural Networks (CNNs) in differentiating between fog and polluted air (smog) in city images through transfer learning. The results demonstrated that Convolutional Neural Networks (CNNs) and Transfer Learning can be used as efficient methods for fog/smog classification.
  • [ X ]
    Öğe
    ESTIMATION OF OCCUPANCY STATUS AND LEVELS FOR INDOOR SPACES
    (Scibulcom Ltd, 2020) Isikdag, U.; Apak, Sudi
    Building energy use today accounts for over 40% of total primary energy consumption. The energy demand for buildings can be decreased through efficient building and facility management. The knowledge related to the use of indoor spaces is a key to successful management. The research aimed to investigate whether the occupancy levels of an indoor space can automatically be determined via machine learning algorithms based on data acquired from multiple indoor sensors. The study involved indoor data collection and a machine learning experiment. The results indicated that machine learning can be considered as a promising approach for the detection of indoor occupancy status and levels.
  • [ X ]
    Öğe
    OPTIMUM MODELLING FOR RECTANGULAR SHAPE REINFORCED CONCRETE (RC) BEAM WITH THE AIM OF MINIMUM CO2 EMISSION
    (Scibulcom Ltd, 2021) Yucel, M.; Bekdas, G.; Isikdag, U.; Apak, Sudi
    Optimum analysis and modelling for any structural member in civil engineering designs are important and they should be also applied and realised by making it possible of providing safe construction, serviceable conditions, and also sustainable requirements. With this aim, in the present study, a simply supported Reinforced concrete (RC) beam with rectangular cross-section was designed with the purpose of reducing and even minimisation of carbon dioxide (CO2) amount emission from structural materials (concrete and steel reinforcement). Furthermore, in the scope of this purpose, optimal beam designs were provided by considering several concrete compressive strength options, and optimal levels for total costs are determined with respect to each design combination, too. To realise this process, the Flower pollination algorithm (FPA) was utilised which is one of the well-known and most-used metaheuristic algorithms. By this means, it can be possible to ensure and generate cost-effective, safe, clean, and sustainable designs by carrying out eco-friendly processes owing to the emission reduction of damaging factors as CO2.