HASNA, SARI PUSPITA (2026) DETEKSI GELEMBUNG PADA FLOW METER BUBBLE MENGGUNAKAN YOLOv8n. S1 thesis, Universitas Mercu Buana Jakarta.
|
Text (HAL COVER)
Cover.pdf Download (821kB) | Preview |
|
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (BAB I)
Bab I.pdf Restricted to Registered users only Download (82kB) |
|
|
Text (BAB II)
Bab II.pdf Restricted to Registered users only Download (238kB) |
|
|
Text (BAB III)
Bab III.pdf Restricted to Registered users only Download (581kB) |
|
|
Text (BAB IV)
Bab IV.pdf Restricted to Registered users only Download (348kB) |
|
|
Text (BAB V)
BAB V.pdf Restricted to Registered users only Download (28kB) |
|
|
Text (DAFTAR PUSTAKA)
Daftar Pustaka.pdf Restricted to Registered users only Download (155kB) |
|
|
Text (LAMPIRAN)
Lampiran.pdf Restricted to Registered users only Download (314kB) |
Abstract
Bubbles, as transparent objects in digital images, present a significant challenge in the field of computer vision, since the minimal color contrast between the object and its background makes accurate edge boundary detection difficult using conventional image processing methods. This research develops a detection, classification, and gas flow rate calculation system based on bubble objects using the YOLOv8n algorithm. The training process was carried out by comparing two pre-processing normalization schemes, namely Grayscale normalization and Coloring normalization (original color), to determine the scheme that yields the best model performance in categorizing bubble OK and bubble NG. The test results show that the model performs better when pre-processed using Coloring normalization, achieving a recall of 98.9%, mAP@50 of 98.4%, and mAP@50-95 of 87.8%. Furthermore, a comparison was conducted using three object tracking methods, namely ByteTrack, OC-Sort, and DeepSort. The test results indicate that the OC-Sort method achieves the highest accuracy, at 90.9%, and was therefore selected as the tracking method for the system. The system was then applied to calculate the gas flow rate based on bubble OK objects passing through two virtual boundary lines, resulting in an average flow rate of 0.917 L/min, with a small number of outlier data caused by ID switching during the tracking process under certain testing conditions. Overall, the system has demonstrated good capability in detecting, classifying, tracking, and applying the results to flow rate calculation using video. Kata kunci: Yolo, bubble, detection, tracking, flowrate, flow meter. Bubble atau gelembung merupakan objek transparan pada citra digital yang menjadi salah satu tantangan dalam bidang computer vision, karena perbedaan kontras warna yang sangat minim antara objek dengan latar belakang yang menyebabkan penetapan batas tepi objek cukup sulit secara akurat oleh proses pengolahan citra digital konvensional. Pada penelitian ini dilakukan pengembangan sistem deteksi, klasifikasi dan pengaplikasian pada perhitungan laju alir gas berbasis gelembung (bubble) menggunakan algoritma YOLOv8n. Proses pelatihan dilakukan dengan membandingkan dua metode normalisasi ketika pre-processing yaitu normalisasi Grayscale dan normalisasi Coloring (mempertahankan warna bawaan) untuk mengetahui performa model terbaik dalam mengategorikan gelembung OK dan gelembung NG. Hasil pengujian menunjukkan performa model lebih baik pada saat dilakukan pre-processing menggunakan normalisasi coloring dengan nilai recall 98,9%, mAP@50 sebesar 98,4% dan mAP50-95 sebesar 87,8%. Selanjutnya dilakukan perbandingan menggunakan 3 metode tracking objek yaitu ByteTrack, OC-Sort, dan Deep Sort. Hasil pengujian menunjukkan metode OC-Sort memiliki tingkat akurasi yang tinggi yaitu 90,9% sehingga dipilih sebagai metode tracking pada sistem. Pengaplikasian sistem kemudian dilakukan untuk menghitung laju aliran gas berdasarkan objek gelembung OK yang melewati dua garis batas virtual dan menghasilkan rata-rata laju aliran sebesar 0,917 L/menit dengan sedikit data outlier yang disebabkan oleh ID switching pada proses tracking pada kondisi pengujian tertentu. Secara keseluruhan, sistem telah menunjukkan kemampuan dengan baik dalam mendeteksi, mengklasifikasi, melacak dan pengaplikasian perhitungan laju aliran menggunakan video. Kata kunci: Yolo, gelembung, deteksi, pelacakan, laju aliran gas, flow meter.
| Item Type: | Thesis (S1) |
|---|---|
| NIM/NIDN Creators: | 41522120016 |
| Uncontrolled Keywords: | Yolo, gelembung, deteksi, pelacakan, laju aliran gas, flow meter. |
| Subjects: | 000 Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 000. Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 004 Data Processing, Computer Science/Pemrosesan Data, Ilmu Komputer, Teknik Informatika 600 Technology/Teknologi > 660 Chemical Engineering and Related Technologies/Teknologi Kimia dan Ilmu yang Berkaitan > 661 Technology of Industrial Chemicals/Teknologi Industri Kimia |
| Divisions: | Fakultas Ilmu Komputer > Informatika |
| Depositing User: | khalimah |
| Date Deposited: | 04 Sep 2026 04:57 |
| Last Modified: | 04 Sep 2026 04:57 |
| URI: | http://repository.mercubuana.ac.id/id/eprint/103622 |
Actions (login required)
 |
View Item |
