OBJECT SEGMENTATION BASED ON THERMAL IMAGES USING DEEP LEARNING (PRE-TRAINED RESNEXT 50)
DOI:
https://doi.org/10.54706/senastindo.v4.2022.207Keywords:
Deep Learning, Fully Convolutional Network, Image Processing, Neural NetworkAbstract
The transportation sector at this time has experienced many technological developments which have been well receifed by the public, especially the people of Indonesia. Along with the development of transportation technology has undergone many developments, with sophistication and increased comfort and better security. So Autonomus Car technology was created that can help drivers to maintain safety while driving. Autonomus car was built using the Neural Network control method, and also Image Processing as signal processing with image input, and with a flip camera used for vehicle input data. Autonomous cars have many positive impacts on human life today, so humans can minimize time properly. Travel safety is maintained, and can be more productive when driving.
The method that is currently developing rapidly is automatic extraction using deep learning. In this final project, automatic extraction method with deep learning technology used is Fully Convolutional Network (FCN) with Residual Neural Network Next (ResNext) architecture.
In this study, the extraction accuracy for automatic vehicle function training reached 98% for ResNext 50 with a resolution of 640x540 pixels. Semantic segmentation will then test with 34030 image frames offline. In ResNext 50 architecture contains 20512 frames in good category, 7883 in adequate category and 5605 in poor category.
Downloads
References
Darenoh, N. V., Prasetyo, Y., Alfarakh, M. G., & Sari, Y. A. (2014). SEGMENTASI SEMANTIK CITRA DENGAN CONVOLUTIONAL NEURAL NETWORK MENGGUNAKAN ARSITEKTUR U-NET.
Pratama, A. B., Kadir, R. E. A., & Hady, M. A. (2022). Deteksi Ruang Kosong pada Jalan Menggunakan Semantic Segmentation pada Mobil Otonom. Jurnal Teknik ITS, 11(1), A23-A28.
Manajang, D., Sompie, S. R. U. A., & Jacobus, A. (2021). Implementasi Framework Tensorflow Object Detection API Dalam Mengklasifikasi Jenis Kendaraan Bermotor. Jurnal Teknik Informatika, 15(3), 171-178.
Wantania, B. B., Sompie, S. R., & Kambey, F. D. (2020). Penerapan Pendeteksian Manusia Dan Objek Dalam Keranjang Belanja Pada Antrian Di Kasir. Jurnal Teknik Informatika, 15(2), 101-108.
Golovanov, S., Kurbanov, R., Artamonov, A., Davydow, A., & Nikolenko, S. (2018). Building detection from satellite imagery using a composite loss function. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (pp. 229-232).
Dahria, M. (2008). Kecerdasan Buatan (Artificial Intelligence). Jurnal Saintikom, 5(2), 185-197.
Samuel, A. L. (1988). Some studies in machine learning using the game of checkers. II—recent progress. Computer Games I, 366-400.
“Fully Connected Layers in Convolutional Neural Networks” https://indiantechwarrior.com/fully-connected-layers-in-convolutional-neural-networks/
“Apa itu Convolutional Neural Networks?” https://medium.com/@16611110/apa-itu-convolutional-neural-network-836f70b193a4
Caroline dkk. (2019). Identifikasi Jalan Kampus Universitas Sriwijaya Berbasis Fully Convolutional Network. SURYA ENERGY, 2(1), 353-358.
Sik-Ho Tsang. (2018). Review: ResNeXt — 1st Runner Up in ILSVRC 2016 (Image Classification). https://towardsdatascience.com/review-resnext-1st-runner-up-of-ilsvrc-2016-image-classification-15d7f17b42ac
“Segmentasi semantik untuk klasifikasi citra.” https://arifiany.medium.com/segmentasi-semantik-untuk-klasifikasi-citra-a004b3906250
NVIDIA Developer. (2018). Jetson AGX Xavier Developer Kit. Diakses pada 16 Januri 2022 dari, https://developer.nvidia.com/embedded/jetson-agx-xavier-developer-kit
Python. (2022). Python. Diakses pada 16 Januari 2022 dari, https://www.python.org/
LabelMe. (2021). LabelMe. Diakses pada 16 Januari 2022 dari, http://LabelMe2.csail.mit.edu/
“FLIR ADK Thermal Vision Automotive Development Kit Teledyne FLIR” https://www.flir.com/products/adk/
“Intersection Over Union” https://medium.com/bisa-ai/intersection-over-union-a8d1532899b3
