Uppsats
The Impact of Non-Domain-Specific Synthetic Data In Automatic Logo Detection : A Unity Perception-Based Study Using YOLO to Train Models with Synthetic Data from Generalized Environments
Master-uppsats
KTH/Skolan för elektroteknik och datavetenskap (EECS)
Publicerad: 2025
Språk: Engelska
Nyckelord
klicka för att sökaSammanfattning
As computer vision becomes increasingly integrated into our society the need to acquire large annotated datasets increases. Synthetic datasets provide a cheaper method of automatically generating annotated data but introduce problems such as domain shift. Furthermore, synthetic dataset are typically domain-specific limiting their reusability across different tasks. This thesis investigates the impact of non-domain-specific synthetic data for training deep learning models in the task of automatic logo detection. Synthetic data was generated using the Unity game engine together with the Unity Perception Package to create varied, annotated images from three general environments: a city, a forest and a room. Using the generated datasets and the real-world Logos in the Wild (LITW) dataset, nine computer vision models were trained utilizing the YOLO11n architecture. These models were evaluated on a LITW and a synthetic dataset and benchmarked against each other. The synthetic models were then fine-tuned with the LITW dataset to explore any further impact on performance. The results showed that models trained solely on non-domain-specific synthetic data performed poorly when benchmarked against the real-world data. However, the fine-tuned models displayed competitive performance, in some cases outperforming the LITW model’s performance especially when trained on a more varied synthetic dataset. These findings suggest that while non-domain-specific synthetic data may not be sufficient in by itself, it can serve as a valuable source in fine-tuning and reduce the need for large amounts of annotated real data.
Information
- Författare
- Shiva Olin, Harald
- Lärosäte / institution
- KTH/Skolan för elektroteknik och datavetenskap (EECS)
- Publiceringsdatum
- 2025
- Uppsatstyp
- Master-uppsats
- Språk
- Engelska
- Nyckelord
- ⌕Artificial Intelligence⌕Machine Learning⌕artificiell intelligens⌕computer vision⌕datorseende⌕Maskininlärning⌕Deep Learning⌕Djupinlärning⌕Supervised Learning⌕synthetic data⌕YOLO⌕Non-Domain-Specific Data⌕Logo Detection⌕Convolutional Neural Networks⌕Unity⌕Unity Perception Package⌕Model evaluation⌕Syntetisk data⌕Icke-Domänspecifik Data⌕Logotypdetektering⌕Faltningsnätvärk⌕modellutvärdering⌕Övervakad Maskininlärning
Utforska vidare
Liknande uppsatser
Uppsatser med liknande ämnen och nyckelord.
Kandidat-uppsats, Högskolan i Halmstad/Akademin för informationsteknologi
Fawal, Raghad
Publicerad: 2026
Master-uppsats, Linnéuniversitetet/Institutionen för management (MAN)
Blbulyan, Erik, Lindhqvist, Hugo
Publicerad: 2026
Yrkesexamen på avancerad nivå, Uppsala universitet/Avdelningen för systemteknik
Vigholm, Albin
Publicerad: 2026
Yrkesexamen på avancerad nivå, Luleå tekniska universitet/Institutionen för ekonomi, teknik, konst och samhälle
Åström, Tuva, Nilsson, Matilda
Publicerad: 2026
Yrkesexamen på avancerad nivå, Luleå tekniska universitet/Institutionen för ekonomi, teknik, konst och samhälle
Nordlander, Jonas
Publicerad: 2026
M1-uppsats, Jönköping University/JTH, Avdelningen för datateknik och informatik
Seyhani Porshekoh, Artin
Publicerad: 2026