Show simple item record

dc.contributor.advisorPriandana, Karlisa
dc.contributor.advisorHardhienata, Medria Kusuma Dewi
dc.contributor.authorRahman, Nabiel Muaafii
dc.date.accessioned2026-08-15T02:40:29Z
dc.date.available2026-08-15T02:40:29Z
dc.date.issued2026
dc.identifier.urihttp://repository.ipb.ac.id/handle/123456789/179210
dc.description.abstractPemantauan dan estimasi jumlah tanaman merupakan aspek penting dalam indoor farming yang memerlukan proses pengawasan secara akurat dan efisien. Penelitian ini bertujuan mengimplementasikan model YOLOv11 untuk mendeteksi tanaman dari video yang diambil menggunakan drone Crazyflie 2.1 berbasis AI-deck serta algoritma ByteTrack untuk melakukan estimasi jumlah tanaman secara otomatis. Proses fine-tuning YOLOv11 menggunakan optimizer AdamW dengan lr0 0,0001, lrf 0,001, momentum 0,9, batch size 1, dan 50 epoch. Selain itu, dilakukan hyperparameter tuning terhadap lima parameter ByteTrack untuk memperoleh konfigurasi terbaik. Data penelitian berupa video jarak 10cm, 30cm, 45cm, 70cm, dan 90cm, lalu diberi anotasi. Hasil evaluasi menunjukkan bahwa model YOLOv11 memperoleh nilai nilai recall sebesar 0,913579, precision sebesar 0,928061 dan mAP50 sebesar 0,928266. Pengujian pada data uji juga menghasilkan precision 0,9324, recall 0,9375, akurasi 0,8785, dan F1-score 0,9349. Pada algoritma ByteTrack, dilakukan pengujian terhadap 56 kombinasi parameter dan diperoleh konfigurasi terbaik dengan nilai sebesar 0,25, rata-rata MOTA 0,78274, dan rata-rata IDF1 0,83932 dalam estimasi jumlah tanaman. Hasil penelitian menunjukkan integrasi YOLOv11 dan ByteTrack mampu membentuk sistem untuk mendeteksi, melacak, dan mengestimasi jumlah tanaman pada lingkungan indoor farming.
dc.description.abstractMonitoring and estimating the number of plants are important aspects of indoor farming that require accurate and efficient monitoring processes. This study aims to implement the YOLOv11 model to detect plants from videos captured using a Crazyflie 2.1 drone equipped with an AI-deck, as well as the ByteTrack algorithm to automatically estimate the number of plants. The YOLOv11 fine-tuning process used the AdamW optimizer with an initial learning rate (lr0) of 0.0001, final learning rate factor (lrf) of 0.001, momentum of 0.9, batch size of 1, and 50 epochs. In addition, hyperparameter tuning was performed on five ByteTrack parameters to obtain the optimal configuration. The research data consisted of videos recorded at distances of 10 cm, 30 cm, 45 cm, 70 cm, and 90 cm, which were subsequently annotated. The evaluation results showed that the YOLOv11 model achieved a recall of 0.913579, precision of 0.928061, and mAP50 of 0.928266. Testing on the test data also resulted in a precision of 0.9324, recall of 0.9375, accuracy of 0.8785, and F1-score of 0.9349. For the ByteTrack algorithm, 56 parameter combinations were evaluated, resulting in the best configuration with a Mean Absolute Error (MAE) of 0.25, an average MOTA of 0.78274, and an average IDF1 of 0.83932 for plant count estimation. The results demonstrate that the integration of YOLOv11 and ByteTrack can form a system capable of detecting, tracking, and estimating the number of plants in an indoor farming environment.
dc.description.sponsorship
dc.language.isoid
dc.publisherIPB Universityid
dc.titleDeteksi dan Estimasi Jumlah Tanaman Menggunakan YOLOv11 dan ByteTrack pada Video Penerbangan Drone Crazyflieid
dc.title.alternativePlant Detection and Counting Using YOLOv11 and ByteTrack on Crazyflie Drone Video Flight
dc.typeSkripsi
dc.subject.keyworddrone crazyflieid
dc.subject.keywordbytetrackid
dc.subject.keywordYOLOv11id
dc.subject.keywordDeteksi Objekid
dc.subject.keywordestimasi jumlah tanamanid
dc.subtypeUndergraduate Theses


Files in this item

Thumbnail
Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record