| fullrecord |
[{"key": "dc.contributor.advisor", "value": "P\u00f6l\u00f6nen, Ilkka Sakari", "language": null, "element": "contributor", "qualifier": "advisor", "schema": "dc"}, {"key": "dc.contributor.author", "value": "Danilova, Ekaterina", "language": null, "element": "contributor", "qualifier": "author", "schema": "dc"}, {"key": "dc.date.accessioned", "value": "2024-06-18T11:06:40Z", "language": null, "element": "date", "qualifier": "accessioned", "schema": "dc"}, {"key": "dc.date.available", "value": "2024-06-18T11:06:40Z", "language": null, "element": "date", "qualifier": "available", "schema": "dc"}, {"key": "dc.date.issued", "value": "2024", "language": null, "element": "date", "qualifier": "issued", "schema": "dc"}, {"key": "dc.identifier.uri", "value": "https://jyx.jyu.fi/handle/123456789/95998", "language": null, "element": "identifier", "qualifier": "uri", "schema": "dc"}, {"key": "dc.description.abstract", "value": "This thesis aims to research and apply various approaches to few-shot object detection to address the real-life problem of detecting bacterial colonies. Bacteria detection is an important topic across multiple fields. Traditional methods include chemical analysis and various image processing algorithms. This thesis investigates the development of the deep learning approach to the selected problem in order to find simple and effective solution.\n\nSince manual data collection is a laborious task, this work puts emphasis on few-shot object detection, focusing on developing an approach optimized for minimal training data with less than 10 labeled images.\n\nVarious techniques, both model- and data-centric, are researched and utilized to identify the most efficient solution. Methods such as data augmentations, meta-learning, and fine-tuning are discussed and evaluated. The main emphasis of the practical part is on exploring the fine-tuning approach, together with data enhancements such as various augmentations and the use of additional close-domain data, and eventually comparing and analyzing their performance.\n\nThe findings suggest that fine-tuning is an effective and simple strategy for few-shot training, particularly when combined with other methods. Incorporating close-domain data significantly enhances model performance, and data augmentations also show good results in expanding the dataset size.\n\nThis work advances the application of few-shot learning in data-constrained environments. Given the specific challenge of bacteria detection, the insights gained are potentially valuable across multiple fields and applications.", "language": "en", "element": "description", "qualifier": "abstract", "schema": "dc"}, {"key": "dc.description.abstract", "value": "T\u00e4m\u00e4 opinn\u00e4ytety\u00f6 pyrkii tutkimaan ja soveltamaan erilaisia l\u00e4hestymistapoja v\u00e4h\u00e4otos-tunnistamiseen bakteeripes\u00e4kkeiden havaitsemisessa.\n\nBakteerien tunnistaminen on t\u00e4rke\u00e4 aihe monilla aloilla. Perinteisi\u00e4 menetelmi\u00e4 ovat kemiallinen analyysi ja erilaiset kuvank\u00e4sittelyalgoritmit. T\u00e4m\u00e4 opinn\u00e4ytety\u00f6 tutkii syv\u00e4oppimisen l\u00e4hestymistapaa valittuun ongelmaan l\u00f6yt\u00e4\u00e4kseen yksinkertaisen ja tehokkaan ratkaisun.\n\nKoska manuaalinen datankeruu on ty\u00f6l\u00e4st\u00e4, t\u00e4ss\u00e4 opinn\u00e4ytety\u00f6ss\u00e4 keskityt\u00e4\u00e4n kehitt\u00e4m\u00e4\u00e4n l\u00e4hestymistapaa v\u00e4h\u00e4otos-tunnistamiseen, joka on optimoitu v\u00e4h\u00e4iselle koulutusdatalle, alle kymmenelle merkitylle kuvalle.\n\nOpinn\u00e4ytety\u00f6ss\u00e4 tutkitaan ja hy\u00f6dynnet\u00e4\u00e4n erilaisia tekniikoita, sek\u00e4 malli- ett\u00e4 datakeskeisi\u00e4, tehokkaimman ratkaisun l\u00f6yt\u00e4miseksi. Siin\u00e4 k\u00e4sitell\u00e4\u00e4n ja arvioidaan menetelmi\u00e4, kuten datan parannusta, metaoppimista ja neuroverkon hienos\u00e4\u00e4t\u00e4mist\u00e4.\n\nK\u00e4yt\u00e4nn\u00f6n osassa p\u00e4\u00e4paino on hienos\u00e4\u00e4t\u00f6menetelmien tutkiminen yhdess\u00e4 datan parannusten, kuten erilaisten hienos\u00e4\u00e4t\u00f6jen ja vastaavanlaisen lis\u00e4datan kanssa ja lopulta niiden suorituskyvyn vertaamisessa ja analysoinnissa.\n\nTulokset osoittavat, ett\u00e4 hienos\u00e4\u00e4t\u00f6 on tehokas ja yksinkertainen strategia v\u00e4h\u00e4otoskoulutukseen, erityisesti kun sit\u00e4 yhdistet\u00e4\u00e4n muihin menetelmiin. Vastaavanlaisen lis\u00e4datan sis\u00e4llytt\u00e4minen parantaa merkitt\u00e4v\u00e4sti mallin suorituskyky\u00e4, ja datan hienos\u00e4\u00e4d\u00f6t osoittavat my\u00f6s hyvi\u00e4 tuloksia tietojoukon laajentamisessa.\n\nT\u00e4m\u00e4 ty\u00f6 edist\u00e4\u00e4 v\u00e4h\u00e4otosoppimisen soveltamista datarajoitteisissa ymp\u00e4rist\u00f6iss\u00e4. Ottaen huomioon bakteerien havaitsemisen haasteellisuuden, saadut tulokset ovat potentiaalisesti arvokkaita useilla aloilla ja sovelluksissa.", "language": "fi", "element": "description", "qualifier": "abstract", "schema": "dc"}, {"key": "dc.description.provenance", "value": "Submitted by jyx lomake-julkaisija (jyx-julkaisija.group@korppi.jyu.fi) on 2024-06-18T11:06:40Z\nNo. of bitstreams: 0", "language": "en", "element": "description", "qualifier": "provenance", "schema": "dc"}, {"key": "dc.description.provenance", "value": "Made available in DSpace on 2024-06-18T11:06:40Z (GMT). No. of bitstreams: 0", "language": "en", "element": "description", "qualifier": "provenance", "schema": "dc"}, {"key": "dc.format.extent", "value": "57", "language": null, "element": "format", "qualifier": "extent", "schema": "dc"}, {"key": "dc.format.mimetype", "value": "application/pdf", "language": null, "element": "format", "qualifier": "mimetype", "schema": "dc"}, {"key": "dc.language.iso", "value": "eng", "language": null, "element": "language", "qualifier": "iso", "schema": "dc"}, {"key": "dc.rights", "value": "CC BY 4.0", "language": "en", "element": "rights", "qualifier": null, "schema": "dc"}, {"key": "dc.title", "value": "Few-shot object detection of bacteria colonies", "language": null, "element": "title", "qualifier": null, "schema": "dc"}, {"key": "dc.type", "value": "master thesis", "language": null, "element": "type", "qualifier": null, "schema": "dc"}, {"key": "dc.identifier.urn", "value": "URN:NBN:fi:jyu-202406184764", "language": null, "element": "identifier", "qualifier": "urn", "schema": "dc"}, {"key": "dc.contributor.faculty", "value": "Faculty of Information Technology", "language": "en", "element": "contributor", "qualifier": "faculty", "schema": "dc"}, {"key": "dc.contributor.faculty", "value": "Informaatioteknologian tiedekunta", "language": "fi", "element": "contributor", "qualifier": "faculty", "schema": "dc"}, {"key": "dc.contributor.organization", "value": "Jyv\u00e4skyl\u00e4n yliopisto", "language": "fi", "element": "contributor", "qualifier": "organization", "schema": "dc"}, {"key": "dc.contributor.organization", "value": "University of Jyv\u00e4skyl\u00e4", "language": "en", "element": "contributor", "qualifier": "organization", "schema": "dc"}, {"key": "dc.type.coar", "value": "http://purl.org/coar/resource_type/c_bdcc", "language": null, "element": "type", "qualifier": "coar", "schema": "dc"}, {"key": "dc.rights.copyright", "value": "\u00a9 The Author(s)", "language": null, "element": "rights", "qualifier": "copyright", "schema": "dc"}, {"key": "dc.rights.accesslevel", "value": "openAccess", "language": null, "element": "rights", "qualifier": "accesslevel", "schema": "dc"}, {"key": "dc.type.publication", "value": "masterThesis", "language": null, "element": "type", "qualifier": "publication", "schema": "dc"}, {"key": "dc.format.content", "value": "fulltext", "language": null, "element": "format", "qualifier": "content", "schema": "dc"}, {"key": "dc.rights.url", "value": "https://creativecommons.org/licenses/by/4.0/", "language": null, "element": "rights", "qualifier": "url", "schema": "dc"}]
|