Sökning: "bilddata"

Visar resultat 11 - 15 av 50 uppsatser innehållade ordet bilddata.

  1. 11. Geospatial Trip Data Generation Using Deep Neural Networks

    Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)

    Författare :Aditya Deepak Udapudi; [2022]
    Nyckelord :Deep Learning; Geospatial; Generative Adversarial Network GAN ; Deep Learning; Geospatial; Generativa Motståndsnätverk GAN ;

    Sammanfattning : Development of deep learning methods is dependent majorly on availability of large amounts of high quality data. To tackle the problem of data scarcity one of the workarounds is to generate synthetic data using deep learning methods. LÄS MER

  3. 12. Exploring Diversity of Spectral Data in Cloud Detection with Machine Learning Methods : Contribution of Near Infrared band in improving cloud detection in winter images

    Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)

    Författare :Nakita Sunil Oza; [2022]
    Nyckelord :Remote sensing; Multispectral imaging; Cloud detection; Data diversity; Deep learning; Fjärranalys; Multispektral bildbehandling; Molndetektion; Datadiversitet; Djupinlärning;

    Sammanfattning : Cloud detection on satellite imagery is an essential pre-processing step for several remote sensing applications. In general, machine learning based methods for cloud detection perform well, especially the ones based on deep learning as they consider both spatial and spectral features of the input image. LÄS MER

  4. 13. Deep Ensembles for Self-Training in NLP

    Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)

    Författare :Axel Alness Borg; [2022]
    Nyckelord :Self-training; Semi-Supervised Learning; Natural Language Processing; Ensembles; Transformers; Knowledge Distillation; Självträning; Semi-Övervakad Inlärning; Datalingvistik; Ensembler; Transformers; Kunskaps Destillering;

    Sammanfattning : With the development of deep learning methods the requirement of having access to large amounts of data has increased. In this study, we have looked at methods for leveraging unlabeled data while only having access to small amounts of labeled data, which is common in real-world scenarios. LÄS MER

  5. 14. Continual Learning and Biomedical Image Data : Attempting to sequentially learn medical imaging datasets using continual learning approaches

    Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)

    Författare :Davit Soselia; [2022]
    Nyckelord :Deep Learning; Continual Learning; Catastrophic Forgetting; Biomedical Image Classification; Djup inlärning; kontinuerligt lärande; katastrofal glömska; biomedicinsk bildklassificering;

    Sammanfattning : While deep learning has proved to be useful in a large variety of tasks, a limitation remains of needing all classes and samples to be present at the training stage in supervised problems. This is a major issue in the field of biomedical imaging since keeping samples in the training sets consistently is often a liability. LÄS MER

  7. 15. Image inpainting methods for elimination of non-anatomical objects in medical images

    Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)

    Författare :Andrea Lorenzo Polo; [2021]
    Nyckelord :Medical Imaging; Tumor Ablation; Inpainting; U-Net; Convolutional Neural Network;

    Sammanfattning : This project studies the removal of non-anatomical objects from medical images. During tumor ablation procedures, the ablation probes appear in the image, hindering the performance of segmentation, registration, and dose computation algorithms. These algorithms can also be affected by artifacts and noise generated by body implants. LÄS MER