ACADSTAFF UGM

CREATION
Title : Investigation of Single Image Depth Prediction under Different Lighting Conditions - A Case Study of Ancient Roman Coins
Author :

Dr. techn. Aufaclav Zatu Kusuma Frisky S.Si., M.Sc. (1) LUKMAN AWALUDIN (2) Prof. Drs. Agus Harjoko, M.Sc., Ph.D. (3) Sebastian Zambanini (4) Robert Sablatnig (5)
Date : 0 2021
Keyword : single image, depth prediction, different lighting conditions, investigation, State-of-theart, Roman Coins. single image, depth prediction, different lighting conditions, investigation, State-of-theart, Roman Coins.
Abstract : This paper investigates the limitations of single image depth prediction (SIDP) under different lighting conditions. Besides that, it also offers a new approach to obtain the ideal condition for SIDP. To satisfy the data requirement, we exploit a photometric dataset, which consists of several images of an object under different light properties. In this case, we used an ancient Roman coin dataset captured under 54 different images with five lighting angles. The ground-truth depth data in the dataset was obtained using the stereo photometric method, and these images are also used as training data in the network. We investigated the capabilities of three different State-of-the-Art (SOTA) methods to reconstruct ancient Roman coins with different lighting scenarios. The first investigation is to compare the performance of a given network using previously trained data to check cross-domains performance. Second, the model is fine-tuned from pre-trained data and trained using 70 percent of the ancient Roman coin dataset. Both models are tested on the rest 30 percent data. As evaluation metrics, RMSE and visual inspection are used. As a result, the methods show different characteristic results based on the lighting condition of the test data. Overall, they perform better at 51° and 71° angles of light so-called ideal condition afterwards. On the other hand, they perform worse at 13° and 32° because of the high density of shadows. They also cannot reach the best performance at 82° caused by the reflection that appears on the image. Based on these findings, we propose a new approach to reduce the shadows and reflections on the image using intrinsic image decomposition to achieve a synthetic ideal condition. Based on the results of synthetic images, this approach can enhance the performance of SIDP. For some SOTA methods, it also achieves better results than previous original RGB images.
Group of Knowledge : Ilmu Komputer
Original Language : English
Level : Internasional
Status :
Published
Document
No Title Document Type Action
1 Investigation of Single Image Depth Prediction under Different Lighting Conditions - A Case Study of Ancient Roman Coins.pdf
Document Type : [PAK] Cek Similarity
[PAK] Cek Similarity View
2 Aufaclav Full Dokumen.pdf
Document Type : [PAK] Full Dokumen
[PAK] Full Dokumen View
3 Korespondensi.pdf
Document Type : [PAK] Bukti Korespondensi Penulis
[PAK] Bukti Korespondensi Penulis View