๋ณธ๋ฌธ ๋ฐ”๋กœ๊ฐ€๊ธฐ

์ „์ฒด ๊ธ€122

[Paper Review] 3D Gaussian Splatting for Real-Time Radiance Field Rendering ์ด ๋…ผ๋ฌธ ๋‚ด์šฉ์œผ๋กœ ๋“ค์–ด๊ฐ€๊ธฐ ์ „์— Pinhole Camera Model ์— ๋Œ€ํ•ด์„œ ์–ด๋Š ์ •๋„ ์ดํ•ด๋ฅผ ํ•˜๊ณ  ๊ฐ€๋ฉด ์ข‹๋‹ค.  Abstract1080p ํ•ด์ƒ๋„์˜ ๋ฌด์ œํ•œ ๋ฐ ์™„์ „ํ•œ ์žฅ๋ฉด์˜ ๋ Œ๋”๋ง์—์„œ๋Š” ์ด์ „๊นŒ์ง€๋Š” ์–ด๋–ค ๋ฐฉ๋ฒ•๋„ ์‹ค์‹œ๊ฐ„ ๋””์Šคํ”Œ๋ ˆ์ด ์†๋„(≥30 fps)๋ฅผ ๋‹ฌ์„ฑํ•˜์ง€ ๋ชปํ•˜๊ณ  ์žˆ๋‹ค.์ด ๋…ผ๋ฌธ์—์„œ๋Š” 1080p ํ•ด์ƒ๋„์—์„œ ๊ณ ํ’ˆ์งˆ ์‹ค์‹œ๊ฐ„(≥30 fps) ์ƒˆ๋กœ์šด ์‹œ์  ํ•ฉ์„ฑ์„ ๊ฐ€๋Šฅํ•˜๊ฒŒ ํ•˜๋Š” ์„ธ ๊ฐ€์ง€ ์ฃผ์š” ์š”์†Œ๋ฅผ ์†Œ๊ฐœํ•œ๋‹ค.1. 3D ๊ฐ€์šฐ์‹œ์•ˆ ํ™œ์šฉ: ์นด๋ฉ”๋ผ ์บ˜๋ฆฌ๋ธŒ๋ ˆ์ด์…˜ ๊ณผ์ •์—์„œ ์ƒ์„ฑ๋œ ํฌ์†Œ ํฌ์ธํŠธ๋ฅผ ์‹œ์ž‘์œผ๋กœ, ์žฅ๋ฉด์„ 3D ๊ฐ€์šฐ์‹œ์•ˆ์œผ๋กœ ํ‘œํ˜„ํ•˜์—ฌ ์—ฐ์†์ ์ธ ๋ณผ๋ฅจ๊ธฐ๋ฐ˜ Radiance Field ์˜ ์žฅ์ ์„ ์œ ์ง€ํ•˜๋ฉด์„œ ๋นˆ ๊ณต๊ฐ„์—์„œ์˜ ๋ถˆํ•„์š”ํ•œ ๊ณ„์‚ฐ์„ ํ”ผํ•œ๋‹ค.2. ์ตœ์ ํ™”์™€ ๋ฐ€๋„ ์ œ์–ด: 3D ๊ฐ€์šฐ์‹œ์•ˆ์˜ ๋น„๋“ฑ๋ฐฉ์„ฑ ๊ณต๋ถ„์‚ฐ์„ ์ตœ์ ํ™”ํ•˜์—ฌ ์žฅ๋ฉด์„ ์ •.. 2024. 12. 14.
[Paper Review] NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis Abstract์ด ๋…ผ๋ฌธ์—์„œ๋Š” ๋ณต์žกํ•œ ์žฅ๋ฉด์˜ ์ƒˆ๋กœ์šด ์‹œ์ ์„ ์ƒ์„ฑํ•˜๋ฉฐ, ๊ธฐ์กด ์—ฐ๊ตฌ๋ฅผ ๋›ฐ์–ด๋„˜๋Š” ์„ฑ๋Šฅ์„ ๋‹ฌ์„ฑํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ์ œ์•ˆํ•œ๋‹ค.์ด ๋ฐฉ๋ฒ•์€ ์†Œ์ˆ˜์˜ ์ž…๋ ฅ ๋ทฐ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์—ฐ์†์ ์ธ volume scene ํ•จ์ˆ˜๋ฅผ ์ตœ์ ํ™”ํ•œ๋‹ค.์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ์žฅ๋ฉด์„ fully connected (not convolution) deep network ๋กœ ํ‘œํ˜„ํ•œ๋‹ค. - ์ž…๋ ฅ: ์—ฐ์†์ ์ธ 5D ์ขŒํ‘œ => ๊ณต๊ฐ„ ์œ„์น˜(x, y, z) ์™€ ์‹œ์  ๋ฐฉํ–ฅ(θ, φ) ์„ ์ž…๋ ฅ- ์ถœ๋ ฅ: ํ•ด๋‹น ์œ„์น˜์˜ ์ฒด์  ๋ฐ€๋„(volume density) ์™€ ์‹œ์  ์ข…์† ๋ฐฉ์ถœ ๋ณต์‚ฌ๊ด‘(view-dependent emitted radiance) ์ถœ๋ ฅ- ํ•ฉ์„ฑ ๊ณผ์ •: ์นด๋ฉ”๋ผ์—์„œ ์žฅ๋ฉด์œผ๋กœ ๋ป—์–ด๋‚˜๊ฐ€๋Š” ๊ด‘์„ (rays) ์„ ๋”ฐ๋ผ 5D ์ขŒํ‘œ(๊ณต๊ฐ„์œ„์น˜์™€ ์‹œ์ ๋ฐฉํ–ฅ)๋ฅผ ์„ ํƒํ•œ ๋’ค, ํ•ด๋‹น ์ขŒํ‘œ์—์„œ ์‹ ๊ฒฝ๋ง์„.. 2024. 12. 10.
[Paper Review] DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation Abstract1. DeepSDF๋ž€?- ํ•™์Šต ๊ธฐ๋ฐ˜์˜ ์—ฐ์†์  Signed Distance Function (SDF) ํ‘œํ˜„- ํ˜•์ƒ ํด๋ž˜์Šค ์ „์ฒด๋ฅผ ๊ณ ํ’ˆ์งˆ๋กœ ํ‘œํ˜„, ๋ณด๊ฐ„(interpolation), ๋ถˆ์™„์ „ ๋ฐ์ดํ„ฐ ๋ณต์› ๊ฐ€๋Šฅ2. ํ‘œํ˜„ ๋ฐฉ์‹- ๋ถ€ํ”ผ ํ•„๋“œ์—์„œ ์ ์˜ ํฌ๊ธฐ: ํ‘œ๋ฉด ๊ฒฝ๊ณ„๊นŒ์ง€์˜ ๊ฑฐ๋ฆฌ- ๋ถ€ํ˜ธ: ํ˜•์ƒ ๋‚ด๋ถ€(-) ๋˜๋Š” ์™ธ๋ถ€(+)- ๊ฒฝ๊ณ„๋Š” ํ•จ์ˆ˜์˜ 0-level-set ์œผ๋กœ ์•”๋ฌต์ ์œผ๋กœ ์ธ์ฝ”๋”ฉ3. ๊ธฐ์กด SDF ์™€ ์ฐจ์ด์ - ๊ธฐ์กด SDF ๋Š” ๋‹จ์ผ ํ˜•์ƒ ํ‘œํ˜„- DeepSDF ๋Š” ํ˜•์ƒ ํด๋ž˜์Šค ์ „์ฒด๋ฅผ ํ•™์Šตํ•˜๊ณ  ํ‘œํ˜„ ๊ฐ€๋Šฅ4. ์„ฑ๊ณผ- 3D ํ˜•์ƒ ํ‘œํ˜„๊ณผ ๋ณต์›์—์„œ ์ตœ์ฒจ๋‹จ ์„ฑ๋Šฅ- ๋ชจ๋ธ ํฌ๊ธฐ๋ฅผ ๊ธฐ์กด ๋Œ€๋น„ 10๋ฐฐ ๊ฐ์†Œ Introduction1. ๋ฌธ์ œ ์ •์˜: 3D ๋”ฅ๋Ÿฌ๋‹ ๋ชจ๋ธ์€ ๊ธฐ์กด ๋ฐฉ์‹์—์„œ ๊ณต๊ฐ„ ๋ฐ ์‹œ๊ฐ„ ๋ณต์žก๋„ ์ฆ๊ฐ€, ์ •์ (vertex) .. 2024. 12. 9.
[Paper Review] DeepVoxels: Learning Persistent 3D Feature Embeddings Abstract์ด ๋…ผ๋ฌธ์€ 3D ์ดํ•ด๊ฐ€ ๋ถ€์กฑํ•œ ์ƒ์„ฑ ์‹ ๊ฒฝ๋ง์˜ ๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•œ ์ƒˆ๋กœ์šด ์ ‘๊ทผ๋ฒ•์„ ์ œ์‹œํ•œ๋‹ค.๊ตฌ์ฒด์ ์œผ๋กœ DeepVoxels ๋Š” 3D ์žฅ๋ฉด์˜ ๋ณต์žกํ•œ ๊ธฐํ•˜ํ•™์„ ๋ชจ๋ธ๋งํ•˜์ง€ ์•Š์œผ๋ฉด์„œ๋„, ์‹œ์ ์— ๋”ฐ๋ผ ๋‹ฌ๋ผ์ง€๋Š” ์™ธํ˜•์„ ์ •ํ™•ํžˆ ์บก์ฒ˜ํ•  ์ˆ˜ ์žˆ๋Š” ํšจ์œจ์ ์ธ 3D ํŠน์ง• ์ž„๋ฒ ๋”ฉ์„ ์ œ์•ˆํ•œ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ์ƒˆ๋กœ์šด ์‹œ์ ์—์„œ ์žฅ๋ฉด์„ ์žฌ๊ตฌ์„ฑํ•˜๋Š” ๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜๊ณ , ๊ธฐ์กด ๋ฐฉ๋ฒ•๋“ค๋ณด๋‹ค ๋” ๋†’์€ ํ’ˆ์งˆ์˜ Synthesis ๊ฒฐ๊ณผ๋ฅผ ์–ป์„ ์ˆ˜ ์žˆ๋‹ค. Introduction Generative Machine Learning:- ์ตœ๊ทผ ๋ช‡ ๋…„ ๋™์•ˆ Generative Machine Learning ๊ธฐ์ˆ ์ด ํฌ๊ฒŒ ๋ฐœ์ „ํ•˜์—ฌ, ์ด๋ฏธ์ง€ ์ƒ์„ฑ๊ณผ ๊ฐ™์€ ์ž‘์—…์—์„œ ๋›ฐ์–ด๋‚œ ์„ฑ๊ณผ๋ฅผ ๋ณด์—ฌ์ฃผ๊ณ  ์žˆ๋‹ค.- ๋ณ€๋ถ„ ์˜คํ† ์ธ์ฝ”๋”(Aariational Autoencoders) ๋‚˜.. 2024. 12. 4.