University of Twente Student Theses
TransNeRF - Improving Neural Radiance fields using transfer learning for efficient scene reconstruction
Paul, Navneet (2021) TransNeRF - Improving Neural Radiance fields using transfer learning for efficient scene reconstruction.
|
PDF
6MB |
| Abstract: | Neural Radiance Fields (NeRF) achieve impressive view synthesis results for a variety of capture settings including forwardfacing capture of bounded and unbounded scenes. We present a transfer learningbased method for neural radiance fields to efficiently synthesize novel views of complex scenes using only a sequence of sample images from the UAVid dataset. We build on transferring the feature color weights of a multilayer perceptron from low resolution images to high resolution scenes by modelling the density and color of the scene as a function of 3D coordinates, latent appearance encodings and viewing directions. Qualitative quantitative metric evaluations are conducted between our proposed approach and other preexisting NeRF and NeRF++ models, delivering a significant improvement over the later in synthesizing novel scenes at high quality and high fps from very less input number of captured images. |
| Item Type: | Essay (Master) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Electrical Engineering MSc (60353) |
| Link to this item: | https://purl.utwente.nl/essays/88717 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page