IDENTIFIKASI DAN KLASIFIKASI VARIETAS TANAMAN HIAS AGLAONEMA BERDASARKAN MORFOLOGI DAUN DENGAN MENGGUNAKAN ALGORITMA YOLOv11

    Diwa Prasetyo, - and Mochamad Iqbal Ardimansyah, - and Raditya Muhammad, - (2025) IDENTIFIKASI DAN KLASIFIKASI VARIETAS TANAMAN HIAS AGLAONEMA BERDASARKAN MORFOLOGI DAUN DENGAN MENGGUNAKAN ALGORITMA YOLOv11. S1 thesis, Universitas Pendidikan Indonesia.

    Abstract

    Identifikasi aglaonema memiliki kesulitan dikarenakan tanaman hias ini memiliki kemiripan antar kultivar yang tinggi. Tantangan identifikasi semakin kompleks karena tingginya kemiripan morfologi antar varietas yakni 0.84 koefisien jaccard. Kesalahan identifikasi pada tanaman aglaonema menjadi tantangan terutama pada pemula. Identifikasi aglaonema umumnya dilakukan secara manual melalui pengamatan visual yang bergantung pada keahlian individu, bersifat repetitif, dan membutuhkan waktu yang lama. Sebelumnya sudah ada penelitian dengan lingkup yang sama dengan menggunakan CNN dan YOLOv8. Namun seiring dengan perkembangan teknologi, algoritma YOLOv11 dikenalkan. Algoritma ini mampu mendeteksi objek secara real-time dengan kecepatan dan akurasi tinggi, sehingga sangat cocok digunakan dalam identifikasi tanaman hias. YOLOv11, menunjukkan peningkatan performa terutama dalam mendeteksi objek berukuran kecil, yang sangat penting untuk membedakan detail morfologi tanaman. Oleh karena itu, penerapan YOLOv11 diharapkan dapat menghasilkan identifikasi Aglaonema yang lebih akurat dan efisien, mengurangi tingkat kesalahan, serta meningkatkan kepercayaan konsumen pada industri tanaman hias. Proses pengembangan model mencakup pengumpulan dataset sebanyak 15 kelas, augmentasi data, pembagian dataset dan tahap modelling. Evaluasi model dilakukan menggunakan mAP, IoU, dan Confusion Matrix untuk memperoleh nilai precision, recall, dan F1-score. Hasil evaluasi menunjukkan kinerja yang memuaskan dengan rata-rata precision 89,9%, recall 93,5%, F1-score 91,7%, mAP50 96%, dan mAP95 81%. Pada tahap pra-implementasi, model mampu melakukan prediksi secara efektif pada skenario satu objek, tanpa objek, maupun multi-objek, yang menunjukkan akurasi deteksi sekaligus ketahanan model terhadap variasi kondisi data. -------- Identification of Aglaonema cultivars is challenging due to the high degree of inter-cultivar similarity. The task becomes even more complex because morphological resemblance across varieties is substantial—i.e., a Jaccard similarity coefficient of 0.84. Misidentification is particularly common among beginners. In practice, identification is typically carried out manually through visual inspection that depends on individual expertise, is repetitive, and is time-consuming. Prior studies in this area have employed CNNs and YOLOv8. With ongoing advances, YOLOv11 has been introduced, offering real-time object detection with high speed and accuracy, making it well suited to ornamental plant identification. YOLOv11 shows improved performance especially for small objects, which is crucial for capturing distinctive morphological details. Accordingly, applying YOLOv11 is expected to yield more accurate and efficient Aglaonema identification, reduce error rates, and enhance consumer confidence in the ornamental plant industry. The model development process encompassed building a 15-class dataset, performing data augmentation, splitting the dataset, and model training. Evaluation used mAP, IoU, and the confusion matrix to derive precision, recall, and F1-score. The results indicate strong performance, with average precision of 89.9%, recall of 93.5%, F1-score of 91.7%, mAP50 of 96%, and mAP95 of 81%. In pre-deployment testing, the model produced effective predictions across single-object, no-object, and multi-object scenarios, demonstrating both accurate detection and robustness to varying data conditions.

    [thumbnail of S_RPL_2103802_Tittle.pdf] Text
    S_RPL_2103802_Tittle.pdf

    Download (604kB)
    [thumbnail of S_RPL_2103802_Chapter1.pdf] Text
    S_RPL_2103802_Chapter1.pdf

    Download (253kB)
    [thumbnail of S_RPL_2103802_Chapter2.pdf] Text
    S_RPL_2103802_Chapter2.pdf
    Restricted to Staf Perpustakaan

    Download (488kB)
    [thumbnail of S_RPL_2103802_Chapter3.pdf] Text
    S_RPL_2103802_Chapter3.pdf

    Download (557kB)
    [thumbnail of S_RPL_2103802_Chapter4.pdf] Text
    S_RPL_2103802_Chapter4.pdf
    Restricted to Staf Perpustakaan

    Download (1MB)
    [thumbnail of S_RPL_2103802_Chapter5.pdf] Text
    S_RPL_2103802_Chapter5.pdf

    Download (245kB)
    [thumbnail of S_RPL_2103802_Appendix.pdf] Text
    S_RPL_2103802_Appendix.pdf
    Restricted to Staf Perpustakaan

    Download (748kB)
    Official URL: https://repository.upi.edu/
    Item Type: Thesis (S1)
    Additional Information: https://scholar.google.com/citations?user=4bmNWxAAAAAJ&hl=en ID SINTA Dosen Pembimbing: Mochamad Iqbal Ardimansyah:6658552 Raditya Muhammad:6682222
    Uncontrolled Keywords: aglaonema, klasifikasi tanaman, Object Recognition, You Only Look Once (YOLO), YOLOv11, plant classification, object recognition
    Subjects: L Education > L Education (General)
    Q Science > QA Mathematics > QA75 Electronic computers. Computer science
    Q Science > QA Mathematics > QA76 Computer software
    Divisions: UPI Kampus cibiru > S1 Rekayasa Perangkaat Lunak
    Depositing User: Diwa Prasetyo
    Date Deposited: 18 Sep 2025 04:30
    Last Modified: 18 Sep 2025 04:30
    URI: http://repository.upi.edu/id/eprint/136722

    Actions (login required)

    View Item View Item