OBJECT SEGMENTATION BASED ON THERMAL IMAGES USING DEEP LEARNING (PRE-TRAINED RESNET 152)
DOI:
https://doi.org/10.54706/senastindo.v4.2022.215Keywords:
Kendaraan Listrik Otonom, Deep Learning, Residual Network 152.Abstract
The latest technological developments in the field of Artificial Intelligence have very rapid capabilities and are able to produce systems that facilitate human activities, especially in the field of transportation, especially driving cars or autonomous electric cars. Artificial Intelligence technology itself is able to support success for object detection by detecting objects using semantic segmentation.
Neural Network and Image processing are methods used to detect objects semantically as input signal processing in the form of images, and the FLIR thermal camera is used as input from the vehicle. The deep learning method uses a Fully Convolutional Network (FCN) with a Residual Network (ResNet) architectural model as its feature extraction. ResNet is an architectural model from FCN that works from this architectural model not to decline even though the architecture is getting deeper, so it can help humans to drive more productively.
The method used in this final project is automatic extraction using deep learning technology with Residual Neural Network 152 (ResNet) architecture. The performance of the semantic segmentation system was tested with 3040 image frames offline using 800 labeled data sets. This method has an extraction accuracy for autonomous vehicle function training reaching 96% with a resolution of 640x512 pixels. The performance of the segmentation system resulted in 18576 image frames in good category, 9333 image frames in sufficient category and 6121 image frames in poor category.
Downloads
References
Amer Sallam. (2021).Early detection of Glaucoma using Transfer Learning from Pre-Trained CNN Models. Diakses pada 08 Agustus 2022 dari, Badan Pusat Statistik (bps.go.id), 2 - 4.
Radhika V M. (2020). Movie Genre Prediction and Recommendation Using Deep Visual Features from Movie Trailers. 3-5.
Agil Bintang Pratama. (2022) Deteksi Ruang Kosong pada Jalan Menggunakan Semantic Segmentation pada Mobil Otonom dari Institut Teknologi Sepuluh Nopember (ITS). 2-3.
Andreanov Ridhovan, dkk (2022) Penerapan Metode Residual Network (RESNET) dalam Klasifikasi Penyakit pada daun Gandum dari Universits Singaperbangsa Karawang. 7-8.
Widi Harsono, Dkk (2021) Klasifikasi Covid-19 Chest X-Ray dengan Tiga Arsitektur CNN (Resnet-152, Inception Resnet-V2, MobileNet-V2). 7-9.
Avneet Pannu. (2015) Artificial Intellegence and Application in Different Areas dari DAV Institute of Engineering and Technology, Jalandhar India. 2-3.
John McCarthy. (1956) The Darmouth Summer Research Project On Artificial Intelligence dari Massacuhetts Institute of Technology. IEE 2-3
Marlina, N. N. A., Ariffin, D. M., Satyawan, A. S., Asysyakuur, M. I., Utamajaya, M. F., Satria, R. A., ... & Ema, E. (2021, December). Sistem Pendeteksi Pejalan Kaki di Lingkungan Terbatas Berbasis SSD Mobilenetv1 Menggunakan Gambar 360 Ternormalisasi. In Prosiding Seminar Nasional Sains Teknologi dan Inovasi Indonesia (SENASTINDO) (Vol. 3, pp. 111-122).
(Pooja Mahajan (2020) Fully Connected vs Convolutional Neural Network)
Caroline dkk. (2019). Identifikasi Jalan Kampus Universitas Sriwijaya Berbasis Fully Convolutional Network. SURYA ENERGY, 2(1), 353-358
(Baishideng. (2021). Infographic With Icons And Timeline For Artificial Intelligence, Machine Learning And Deep Learning.
(Lina, Q. (2019) Apa Itu Convolutional Neural Network)
He dkk. (2016). Deep Residual Learning For Image Recognition. Proceeding of the IEEE Conference Computer Vision and Pattern Recognition, 770-778.
Aditama, Nadia. (2020). Segmentasi Citra Berbasis Deep Learning https://informatika.stei.itb.ac.id/~rinaldi.munir/Citra/2021-2022/27-Kuliah-tamu-1.pdf
NVIDIA Developer. (2018). Jetson AGX Xavier Developer Kit. Diakses pada 16 Januri 2022 dari, https://developer.nvidia.com/embedded/jetson-agx-xavier-developer-kit
NVIDIA Developer. (2018). Jetson AGX Xavier Developer Kit. Diakses pada 16 Januri 2022 dari, https://developer.nvidia.com/embedded/jetson-agx-xavier-developer-kit
LabelMe. (2021). LabelMe. Diakses pada 16 Januari 2022 dari, http://LabelMe2.csail.mit.edu/
FLIR ADK Thermal Vision Automotive Development Kit Teledyne FLIR
