Efy Yosrita, Efy
Unknown Affiliation

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

PERBAIKAN INTERPOLASI GERAKKAN MODEL SKELETON 3D DARI DATASET HASIL KAMERA KINECT Rusjdi, Darma; Wulandari, Dewi Arianti; Yosrita, Efy
Seminar Nasional Informatika (SEMNASIF) Vol 1, No 1 (2020): Peran Digital Society dalam Pemulihan Pasca Pandemi
Publisher : Jurusan Teknik Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

AbstractThe Kinect camera's dataset can be used for training and testing of human movement recognition using a deep learning approach, in addition to tracking and estimating the position and position of the human body. Improvement of human movement which is covered by other body parts is still a challenge. The research objective: to design a repair application to move the 3D skeleton model from the Kinect camera dataset using the 3D interpolation histogram smoothing approach on the human body. The study was in the form of a simulation of a skeleton model movement improvement in the context of developing a key-frame based animation application through Kinect camera recording. The prototype development method goes through the cycle stages of analysis, design, implementation. The initial analysis stage selects the kinect camera dataset by examining the file format and data structure. Furthermore, the development for data improvement through the 3D interpolation refinement approach. Movement improvement in terms of measurement of interpolated resultant errors from each time t0, t1, t2 and t3 histogram using the RMSE method and visual observation. showed significant results close to normal movement.Keywords : dataset, kinect camera, interpolation, skeleton modelDataset hasil kamera Kinect dapat digunakan untuk pelatihan dan pengujian pengenalan gerakkan manusia dengan pendekatan deep learning, selain tracking dan estimasi letak dan posisi tubuh manusia. Perbaikan gerakkan manusia yang tertutup bagian tubuh lainnya masih menjadi tantangan. Tujuan penelitian: membuat rancang bangun aplikasi perbaikan gerakkan model skeleton 3D dari dataset hasil kamera Kinect menggunakan pendekatan penghalusan histogram interpolasi 3D pada bagian tubuh manusia. Kajian berupa simulasi perbaikan gerakkan model skeleton dalam rangka pengembangan aplikasi animasi berbasis key-frame melalui perekaman kamera Kinect. Metode pengembangan prototitpe melalui tahapan siklus analisis, disain, implementasi. Tahap analisis awal memilih dataset hasil kamera kinect dengan mengkaji format file dan struktur data. Kemudian mengekstraksi beberapa kategori gerakkan kedalam kedalam format file untuk eksperimen. Melalui disain awal pengembangan program dilakukan penyesuaian skala angka untuk menghasilkan histogram yang dapat memperlihatkan bagian kesalahan gerakkan secara signikan menggunakan Root Mean Squared Error (RMSE). Selanjutnya pengembangan untuk perbaikan data melalui pendekatan penghalusan interpolasi 3D. Perbaikan gerakkan ditinjau dari pengukuran kesalahan resultante interpolasi dari setiap watu t0, t1, t2 dan t3 histogram dengan metode RMSE dan pengamatan secara visual. menunjukkan hasil cukup signifikan mendekati gerakkan yang normal.Kata Kunci : dataset, kamera kinect, interpolasi, model skeleton
A Single Image Dehazing using U-Net and Lightweight Vision Transformer Aditya, Dion; Yosrita, Efy
Jurnal E-Komtek (Elektro-Komputer-Teknik) Vol 9 No 1 (2025)
Publisher : Politeknik Piksi Ganesha Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37339/e-komtek.v9i1.2352

Abstract

This research presents a single-image dehazing method that integrates a Lightweight Vision Transformer (LVT) and U-Net to capture both local and global features. LVT enhances resolution, U-Net extracts local features, and LVT refines global dependencies before fusion. Evaluations on O-Haze and HSTS datasets show PSNR scores of 27.88 (O-Haze, ResNet-50) and 28.22 (HSTS, no backbone), outperforming existing methods while maintaining competitive SSIM. The results demonstrate effectiveness in real-world haze scenarios, such as wildfire-induced haze in Indonesia.