Sumbogo, Jajar and -, Gunawan Ariyanto, S.T., M.Comp.Sc, Ph.D (2026) Analisis Sensitivitas Performa Model YOLOv11 Terhadap Kualitas Citra Input dalam Sistem Deteksi Keterisian Tempat Parkir. Skripsi thesis, Universitas Muhammadiyah Surakarta.
![[img]](https://eprints.ums.ac.id/style/images/fileicons/application_pdf.png) |
PDF (Naskah Publikasi)
NASKAH PUBLIKASI.pdf Download (1MB) |
|
PDF (Surat Pernyataan Publikasi)
SURAT PERNYATAAN.pdf Restricted to Repository staff only Download (3MB) |
Abstract
The implementation of smart parking systems in urban environments is often hampered by non-ideal real-world image conditions, such as low resolution, poor lighting, and occlusion. The reliability of parking occupancy classification systems is highly dependent on the performance of Computer Vision models in the face of such image quality degradation, but quantitative understanding of this performance sensitivity is still limited. Therefore, this study aims to quantitatively analyze the relationship between increasing vehicle counting errors as measured by Mean Absolute Error (MAE) and decreasing occupancy classification accuracy, in order to identify the critical point at which the system is no longer reliable. The method used is a quantitative experiment with an image degradation simulation approach. A YOLOv11 model-based inference system in ONNX (Open Neural Network Exchange) format is tested on various image quality scenarios, where performance analysis is performed at two levels: (1) counting errors as measured by MAE and Root Mean Square Error (RMSE), and (2) final classification accuracy as measured by Classification Accuracy and Confusion Matrix. The novelty of this study lies in the two-level performance analysis that explicitly links failures at the detection-count level with failures at the system level. This research produces a series of graphs that map the relationship between the level of degradation with MAE and classification accuracy, and identifies an MAE threshold value (critical point) that can be used as a reference for the operational feasibility of the system in the field.
| Item Type: | Thesis (Skripsi) |
|---|---|
| Uncontrolled Keywords: | Smart Parking, YOLO, Occupancy Classification, Sensitivity Analysis, Image Quality, Computer Vision, MLOps |
| Subjects: | T Technology > TE Highway engineering. Roads and pavements T Technology > TL Motor vehicles. Aeronautics. Astronautics T Technology > Technical Information T Technology > Technical Information > Software. Aplication > Pemograman |
| Divisions: | Fakultas Komunikasi dan Informatika > S1 Teknik Informatika |
| Depositing User: | JAJAR SUMBOGO |
| Date Deposited: | 23 Feb 2026 04:48 |
| Last Modified: | 23 Feb 2026 04:48 |
| URI: | http://eprints.ums.ac.id/id/eprint/143278 |
Actions (login required)
 |
View Item |
