TY - THES A1 - Antonius Didi Kurniadi, - A1 - Munawir, - A1 - Muhammad Taufik Dwi Putra, - Y1 - 2025/08/19/ UR - https://repository.upi.edu/ N1 - https://scholar.google.com/citations?hl=en&user=RZ3a3lMAAAAJ&view_op=list_works&authuser=4&gmla=AH8HC4y896o845-VMltZJZ__X5-HinEcqPMLhVoDlcl3EyI5fN4riVe8MI15pcJmSS-alCtDzVz7DEsRlgwgVPESRTpyoevwE5rsCzJ7OErUOLUfutEN4NMspC7cVQNTpNM ID SINTA Dosen Pemimbing: Munawir: 6745899 Muhammad Taufik Dwi Putra: 6745726 TI - DETEKSI DAN PEMETAAN SAMPAH PLASTIK DI SUNGAI MENGGUNAKAN CITRA DRONE DAN YOLOV11-N BERBASIS WEBSITE AV - restricted N2 - Pencemaran sungai oleh sampah plastik merupakan masalah lingkungan yang mendesak dan memerlukan penanganan berkelanjutan. Pemantauan langsung sering terkendala karena luasnya area sungai yang harus diawasi. Penelitian ini mengembangkan sistem deteksi dan pemetaan sampah plastik yang mengapung di sungai menggunakan citra drone dan model deteksi objek YOLOv11-n. Sistem dirancang mengenali lima jenis sampah plastik, yaitu kantong plastik, botol plastik, kemasan plastik, gelas plastik, dan styrofoam, serta memperkirakan lokasi objek sampah yang terdeteksi dengan mengonversi posisi piksel objek sampah ke koordinat geografis menggunakan pendekatan Ground Sampling Distance (GSD), dengan titik tengah citra sebagai acuan. Uji lapangan pada dua lokasi memperlihatkan hasil yang bervariasi. Pada lokasi pertama, performa deteksi relatif stabil dan mendekati hasil pengamatan manual. Pada lokasi kedua akurasi lebih rendah akibat intensitas cahaya yang tinggi sehingga memicu kesalahan klasifikasi, terutama antar objek dengan kemiripan warna. Faktor lain yang memengaruhi akurasi meliputi ukuran objek kecil pada ketinggian tinggi, penurunan resolusi citra, serta keterbatasan variasi dataset. Model menunjukkan performa baik dengan precision 0,827, recall 0,711, F1-score 0,765, mAP@0.5 sebesar 0,813, dan mAP@0.5:0.95 sebesar 0,535. Sistem yang dirancang telah berhasil diimplementasikan dalam bentuk website deteksi sampah plastik serta dengan pendekatan GSD, posisi sampah yang terdeteksi dalam satuan piksel pada citra drone diubah menjadi jarak nyata. -------- Plastic waste pollution in rivers is an urgent environmental problem that requires sustainable solutions. Direct monitoring is often constrained by the vast areas of rivers that need to be observed. This study developed a system for detecting and mapping floating plastic waste in rivers using drone imagery and the YOLOv11-n object detection model. The system is designed to recognize five types of plastic waste, namely plastic bags, plastic bottles, plastic packaging, plastic cups, and styrofoam, as well as to estimate the locations of detected waste objects by converting their pixel positions into geographic coordinates using the Ground Sampling Distance (GSD) approach, with the image center as a reference point. Field tests conducted at two locations showed varying results. At the first location, detection performance was relatively stable and closely aligned with manual observations. At the second location, accuracy was lower due to high light intensity, which triggered misclassifications, particularly among objects with similar colors. Other factors affecting accuracy included small object sizes at higher altitudes, reduced image resolution, and limited dataset variation. The model demonstrated strong performance with a precision of 0.827, recall of 0.711, F1-score of 0.765, mAP@0.5 of 0.813, and mAP@0.5:0.95 of 0.535. The system was successfully implemented as a plastic waste detection website, and with the GSD approach, the pixel-based positions of detected waste in drone imagery were converted into real-world distances. PB - Universitas Pendidikan Indonesia M1 - other ID - repoupi137335 KW - Deteksi Objek KW - Pemetaan Sampah KW - Sampah Plastik KW - Citra Drone KW - YOLOv11-n KW - Object Detection KW - Waste Mapping KW - Plastic Waste KW - Drone Imagery ER -