Sistem peringatan bagi penyandang tunanetra terhadap objek bergerak melalui mobile camera = Warning system for visually impaired people towards moving objects through mobile camera

Aldi, Sebastian (2020) Sistem peringatan bagi penyandang tunanetra terhadap objek bergerak melalui mobile camera = Warning system for visually impaired people towards moving objects through mobile camera. Bachelor thesis, Universitas Pelita Harapan.

	Text (Title) Title.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)
Preview	Text (Abstract) Abstract.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (166kB) | Preview
Preview	Text (ToC) ToC.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (217kB) | Preview
Preview	Text (Chapter1) Chapter1.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (178kB) | Preview
	Text (Chapter2) Chapter2.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (749kB)
	Text (Chapter3) Chapter3.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (522kB)
	Text (Chapter4) Chapter4.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (709kB)
	Text (Chapter5) Chapter5.pdf Restricted to Registered users only Available under License Creative Commons Attribution Share Alike. Download (95kB)
Preview	Text (Bibliography) Bibliography.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (169kB) | Preview
	Text (Appendices) Appendices.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)

Abstract

Menurut laporan dari World Health Organization, ada sekitar satu milyar penyandang tunanetra di seluruh dunia. Melalui telepon genggam, penyandang tunanetra dapat memakai aplikasi navigasi seperti Google Maps atau Waze untuk mencapai suatu tujuan. Namun, aplikasi-aplikasi tersebut tidak menyediakan sarana untuk pengenalan objek bergerak. Oleh karena itu, penulis merancang sistem berupa aplikasi Android dengan nama Blindness Guidance untuk mendeteksi objek bergerak melalui kamera telepon genggam. Dengan adanya sistem ini, pengguna dapat menerima peringatan secara waktu nyata mengenai adanya keberadaan objek-objek bergerak di depannya agar terhindar dari kecelakaan, terutama di tempat yang padat pejalan kaki. Blindness Guidance dimulai dengan tahap pemuatan aset, yaitu model MobileNets dan daftar label agar TensorFlow Lite dapat mengenali objek dari citra yang disediakan. Kemudian, API Camera2 menyediakan citra yang tertangkap oleh kamera telepon genggam. Citra yang ditangkap kemudian diproses oleh sistem menjadi deretan angka yang menjadi input bagi TensorFlow Lite. Terakhir, Output dari TensorFlow Lite yang merupakan nama dan lokasi objek pada citra kemudian dipakai dalam kalkulasi jarak dari objek ke pengguna. Pengujian sistem terbagi menjadi dua bagian, yaitu jarak dan akurasi. Uji jarak dilakukan melalui pemakaian tripod dan meteran. Jarak asli akan dibandingkan dengan jarak yang terhitung oleh sistem. Uji akurasi dilakukan melalui analisis rekaman dua menit dari sistem, yang disusul dengan perhitungan precision dan recall. Menurut hasil pengujian, jarak yang terkalkulasi memiliki galat (error) margin di bawah 5% untuk jarak di atas empat meter. Untuk hasil pengujian akurasi, nilai mean average precision adalah 0.9393, sementara nilai mean average recall adalah 0.4479. Hasil ini menunjukkan bahwa sistem dapat mendeteksi objek berjarak di bawah delapan meter. / According to a report by the World Health Organization, there are about one billion visually impaired people in the world. By using their phones, visually impaired people can utilize navigation applications such as Google Maps and Waze to arrive at their destination. However, those applications do not provide the means to detect moving objects. Therefore, the writer designed an Android application called Blindness Guidance to detect moving objects by using the camera of their smartphone. With the presence of such application, users can be warned in real-time about surrounding objects in front of them to avoid accidents, especially in highly populated areas. Blindness Guidance starts by loading the assets that are needed, consisting of the model file and the labels of the detected object. These assets are required for TensorFlow Lite to detect and classify objects from images. Next, the camera provides the images required in real time. The captured images are then processed into an array of numbers that are treated as input for TensorFlow Lite. Lastly, the output from TensorFlow Lite which consists of the name and location of the object in the image is then used to calculate the distance from the object to the user. System testing is divided into two parts, which are distance and accuracy. Distance testing is done by using a tripod and tape measure. The real distance is then compared to the calculated distance. Accuracy testing is done by analyzing a two-minute recording, followed by calculating the precision and recall. According to the testing, the calculated distance has an error margin of below 5% for distances above four meters. For the accuracy testing, the mean average precision is 0.9393, while the mean average recall is 0.4479. The results show that objects below eight meters are successfully detected.

Actions (login required)

View Item