Chalmers Open Digital Repository

Välkommen till Chalmers öppna digitala arkiv!

Här hittar du:

  • Studentarbeten utgivna på lärosätet, såväl kandidatarbeten som examensarbeten på grund- och masternivå
  • Digitala specialsamlingar, som t ex Chalmers modellkammare
  • Utvalda projektrapporter
 

Enheter i Chalmers ODR

Välj en enhet för att se alla samlingar.

Senast inlagda

Self-Supervised Fixed-Scene Adaptation for Object-Detection in Real-Time Surveillance: A Comparative Study of YOLO11 and RF-DETR
(2026) Justad, Jacob
This thesis investigates self-supervised fixed-scene adaptation for real-time objectdetectors in an edge-computing surveillance context. While modern object-detectors achieve strong results on general-purpose benchmarks, deployment in static camera scenes introduces distinct challenges: domain shift to a specific viewpoint, limited availability of scene-specific labels, and stringent compute and memory budgets ondevice. At the same time, the stationary background of surveillance footage provides exploitable structures, as do their temporal dependencies of video-frames. This study conducts a comparative analysis of two state-of-the-art object detection architectures: the Transformer-dominant RF-DETR and the convolutional neural network (CNN)- dominant YOLO11. The thesis employs the 100Scenes dataset to represent a broad range of surveillance environments. Experimental results demonstrate that RF-DETR consistently achieves higher accuracy, smoother convergence, and greater robustness than YOLO11, albeit with higher hardware demands. In contrast, YOLO11 variants (with a frozen backbone) leverage the larger trainable capacity of the neck and head to enable high scene-specific adaptability. While this yields significant gains under quality labeling, it tends to increase sensitivity to imperfect pseudo-labels and the risk of overfitting. Furthermore, by systematically varying model scales, adaptation strategies and environmental conditions the experimental design yields more than 3400 distinct runs. First the work examines the extent to which smaller, specialized models can match the approach of substantially larger models. The experimental results show that a small specialised model can compete with larger general models. Secondly, the study evaluated a proposed on-device self-supervised labeling strategy that integrates SAHI with a bidirectional implementation of ByteTrack. The proposed self-supervised labeling strategy provided reliable performance gains across all architectures and configurations, by recovering hard negatives, more specifically small, occluded and low confidence instances. Thirdly, the study investigated background-context fusion (BF). It proved to be consistently improving the performance in general for RFDETR, while it proved inconsistent for YOLO11 and failed to increase robustness against seasonality, suggesting it induced background-dependent overfitting. Finally, the study shows that all models being trained on a summer scene exhibit a decrease in relative performance compared with the non-adapted models during a seasonal domain shift to a winter scene.
Budget efficient 3D-mesh of engravings using photometric stereo
(2025) Heinmetz, Edvard
This paper describes a process for calculating approximate light positions in a photo set captured of an inscribed wall, with the purpose of applying photometric stereo. The light sources across the photo set are of constant intensity and their positions are inferred using one reflective indicator sphere and the lighting of the wall. The light positions are then used to create a depth map of the wall in an optimization scheme that alternates between depth and albedo values. Various strategies for using the depth map to highlight and isolate inscriptions are then applied.
Designing a Collaboration and Communication Process for a Garment Renewal Service
(2026) Gunnelius, Lovisa
The textile industry has a significant environmental impact and is one of the largest contributors to land and water use, greenhouse gas emissions, and raw material consumption. Transitioning the industry toward a circular economy is therefore essential, and one way to support this shift is by extending the lifespan of garments through repair and refresh. RecoMended is a company that provides industrial scale garment renewal services for other companies, enabling the resale or continued use of existing products. However, as industrial scale garment renewal services are new to the market, established collaboration and communication processes for developing service specifications are limited. As a result, the collaboration and communication process currently used by RecoMended is time-consuming, resource-intensive, and highly customerspecific, making it difficult to scale. This study aims to investigate how the collaboration and communication process between RecoMended and its customers can be improved to support more efficient development of service specifications for garment renewal and allow RecoMended to scale its production. The research was conducted through interviews, workshops, and contextual inquiries with RecoMended and several of its customer companies. The collected data were analyzed and synthesized, resulting in the design of a proposed future collaboration and communication process together with a service toolkit. The proposed design outlines a structured, step-based collaboration and communication process that is intended to support RecoMended and its customers in developing service specifications in a more informed and efficient way. The process is envisioned to be supported by a service toolkit consisting of standardized service packages intended to enable scalability in the production workshop, a reference library with visual and material examples of available renewal procedures, and a set of guiding questions designed to ensure that key decisions are addressed. While further development and testing would be required, the proposed process and service toolkit are intended to provide a foundation for clearer communication, more informed decision-making, and improved collaboration, thereby supporting RecoMended’s ability to scale its garment renewal services.
Exploring Image-to-Text Visual Search Using Open Source Models
(2026) Liu, Tommy
Visual searching refers to the use of visual data, typically images, in order to perform a search rather than textual input. Most visual search implementations rely on performing similarity searching over image features, in which a user-submitted query image is compared against all searchable entries’ features before returning sufficiently similar results. This thesis explores a different method which utilizes image descriptions generated by vision-language models instead of image features, where the descriptions are converted into embeddings in order to match with other search entries. Evaluation data indicate that the method can provide satisfactory retrieval performance in addition to maintaining a low search query execution time, provided that an adequate vision-language model is employed and sufficient server capacity is available.