๋ณธ๋ฌธ ๋ฐ”๋กœ๊ฐ€๊ธฐ
728x90
๋ฐ˜์‘ํ˜•

3D9

NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis ๐Ÿ’ก 0. Abstract ์šฐ๋ฆฌ๋Š” ๋“œ๋ฌธ ์ž…๋ ฅ ๋ทฐ ์„ธํŠธ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์—ฐ์†์ ์ธ ๋ถ€ํ”ผ ์žฅ๋ฉด ํ•จ์ˆ˜๋ฅผ ์ตœ์ ํ™”ํ•˜์—ฌ ๋ณต์žกํ•œ ์žฅ๋ฉด์˜ ์ƒˆ๋กœ์šด ์‹œ์ ์„ ํ•ฉ์„ฑํ•˜๋Š” ์ตœ์ฒจ๋‹จ ๊ฒฐ๊ณผ๋ฅผ ๋‹ฌ์„ฑํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ์ œ์‹œํ•ฉ๋‹ˆ๋‹ค. ์šฐ๋ฆฌ์˜ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ์™„์ „ํžˆ ์—ฐ๊ฒฐ๋œ (๋น„์„ ํ˜•) ์‹ฌ์ธต ๋„คํŠธ์›Œํฌ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์žฅ๋ฉด์„ ํ‘œํ˜„ํ•˜๋ฉฐ, ์ž…๋ ฅ์€ ๋‹จ์ผ ์—ฐ์†์ ์ธ 5D ์ขŒํ‘œ (๊ณต๊ฐ„ ์œ„์น˜ (x, y, z) ๋ฐ ์‹œ์ฒญ ๋ฐฉํ–ฅ (θ, φ))์ด๊ณ  ์ถœ๋ ฅ์€ ํ•ด๋‹น ๊ณต๊ฐ„ ์œ„์น˜์—์„œ์˜ ๋ถ€ํ”ผ ๋ฐ€๋„์™€ ์‹œ์ ์— ์˜์กดํ•˜๋Š” ๋ฐฉ์ถœ ๋ž˜๋””์–ธ์Šค์ž…๋‹ˆ๋‹ค. ์šฐ๋ฆฌ๋Š” ์นด๋ฉ”๋ผ ๊ด‘์„ ์„ ๋”ฐ๋ผ 5D ์ขŒํ‘œ๋ฅผ ์ฟผ๋ฆฌํ•˜์—ฌ ๋ทฐ๋ฅผ ํ•ฉ์„ฑํ•˜๊ณ , ์ „ํ†ต์ ์ธ ๋ถ€ํ”ผ ๋ Œ๋”๋ง ๊ธฐ์ˆ ์„ ์‚ฌ์šฉํ•˜์—ฌ ์ถœ๋ ฅ ์ƒ‰์ƒ๊ณผ ๋ฐ€๋„๋ฅผ ์ด๋ฏธ์ง€๋กœ ํˆฌ์˜ํ•ฉ๋‹ˆ๋‹ค. ๋ถ€ํ”ผ ๋ Œ๋”๋ง์€ ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ๋ฏธ๋ถ„ ๊ฐ€๋Šฅํ•˜๊ธฐ ๋•Œ๋ฌธ์—, ์šฐ๋ฆฌ์˜ ํ‘œํ˜„์„ ์ตœ์ ํ™”ํ•˜๊ธฐ ์œ„ํ•ด ํ•„์š”ํ•œ ์œ ์ผํ•œ ์ž…๋ ฅ์€ ์•Œ๋ ค์ง„ ์นด๋ฉ”๋ผ ํฌ์ฆˆ๋ฅผ ๊ฐ€์ง„ ์ด.. 2023. 7. 13.
[1์ฃผ์ฐจ] NeRF: Representing Scenes asNeural Radiance Fields for View Synthesis ๐Ÿ’ก 0. Abstract ์šฐ๋ฆฌ๋Š” ๋“œ๋ฌธ ์ž…๋ ฅ ๋ทฐ ์„ธํŠธ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์—ฐ์†์ ์ธ ๋ถ€ํ”ผ ์žฅ๋ฉด ํ•จ์ˆ˜๋ฅผ ์ตœ์ ํ™”ํ•˜์—ฌ ๋ณต์žกํ•œ ์žฅ๋ฉด์˜ ์ƒˆ๋กœ์šด ์‹œ์ ์„ ํ•ฉ์„ฑํ•˜๋Š” ์ตœ์ฒจ๋‹จ ๊ฒฐ๊ณผ๋ฅผ ๋‹ฌ์„ฑํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ์ œ์‹œํ•ฉ๋‹ˆ๋‹ค. ์šฐ๋ฆฌ์˜ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ์™„์ „ํžˆ ์—ฐ๊ฒฐ๋œ (๋น„์„ ํ˜•) ์‹ฌ์ธต ๋„คํŠธ์›Œํฌ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์žฅ๋ฉด์„ ํ‘œํ˜„ํ•˜๋ฉฐ, ์ž…๋ ฅ์€ ๋‹จ์ผ ์—ฐ์†์ ์ธ 5D ์ขŒํ‘œ (๊ณต๊ฐ„ ์œ„์น˜ (x, y, z) ๋ฐ ์‹œ์ฒญ ๋ฐฉํ–ฅ (θ, φ))์ด๊ณ  ์ถœ๋ ฅ์€ ํ•ด๋‹น ๊ณต๊ฐ„ ์œ„์น˜์—์„œ์˜ ๋ถ€ํ”ผ ๋ฐ€๋„์™€ ์‹œ์ ์— ์˜์กดํ•˜๋Š” ๋ฐฉ์ถœ ๋ž˜๋””์–ธ์Šค์ž…๋‹ˆ๋‹ค. ์šฐ๋ฆฌ๋Š” ์นด๋ฉ”๋ผ ๊ด‘์„ ์„ ๋”ฐ๋ผ 5D ์ขŒํ‘œ๋ฅผ ์ฟผ๋ฆฌํ•˜์—ฌ ๋ทฐ๋ฅผ ํ•ฉ์„ฑํ•˜๊ณ , ์ „ํ†ต์ ์ธ ๋ถ€ํ”ผ ๋ Œ๋”๋ง ๊ธฐ์ˆ ์„ ์‚ฌ์šฉํ•˜์—ฌ ์ถœ๋ ฅ ์ƒ‰์ƒ๊ณผ ๋ฐ€๋„๋ฅผ ์ด๋ฏธ์ง€๋กœ ํˆฌ์˜ํ•ฉ๋‹ˆ๋‹ค. ๋ถ€ํ”ผ ๋ Œ๋”๋ง์€ ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ๋ฏธ๋ถ„ ๊ฐ€๋Šฅํ•˜๊ธฐ ๋•Œ๋ฌธ์—, ์šฐ๋ฆฌ์˜ ํ‘œํ˜„์„ ์ตœ์ ํ™”ํ•˜๊ธฐ ์œ„ํ•ด ํ•„์š”ํ•œ ์œ ์ผํ•œ ์ž…๋ ฅ์€ ์•Œ๋ ค์ง„ ์นด๋ฉ”๋ผ ํฌ์ฆˆ๋ฅผ ๊ฐ€์ง„ ์ด.. 2023. 7. 13.
[1์ฃผ์ฐจ] [EECS 498-007 / 598-005] 3D Vision ๊ฐ•์˜ ์ •๋ฆฌ 1. 3D Vision Topics 2. 3D Shape Representations 2.1 3D Shape Representations: Depth Map ๐Ÿ’ก ํ”ฝ์…€์— ๋Œ€ํ•ด ์นด๋ฉ”๋ผ์™€ ํ”ฝ์…€์˜ ๊ฑฐ๋ฆฌ๋ฅผ ๊ตฌํ•˜๋Š” ๋ฐฉ์‹ + ์‹œ์•ผ์˜ ๋‹จ์ ์„ ๋ณด์™„ํ•˜๊ธฐ ์œ„ํ•œ loss ๊ตฌ๋น„ https://arxiv.org/abs/1411.4734 Depth map์€ ๊ฐ pixel์— ๋Œ€ํ•ด ์นด๋ฉ”๋ผ์™€ ํ”ฝ์…€ ์‚ฌ์ด์˜ ๊ฑฐ๋ฆฌ๋ฅผ ๊ตฌํ•จ ๊ธฐ์กด segmentation ์ฒ˜๋Ÿผ FC๋ฅผ ํ†ตํ•ด pixel ๋ณ„๋กœ ๊ณ„์‚ฐํ•  ์ˆ˜ ์žˆ๋‹ค ์ด๋ฅผ ํ†ตํ•ด Predicted Depth Image์™€ Ground-Truth image์™€์˜ Per-pixel loss๋ฅผ ๊ณ„์‚ฐํ•˜์—ฌ ํ•™์Šตํ•  ์ˆ˜ ์žˆ์Œ ํ•˜์ง€๋งŒ ์šฐ๋ฆฌ ๋ˆˆ์—์„œ๋Š” ์ž‘๊ณ  ๊ฐ€๊นŒ์ด ์žˆ๋Š” ๋ฌผ์ฒด์™€ ํฌ๊ณ  ๋ฉ€๋ฆฌ ์žˆ๋Š” ๋ฌผ์ฒด์˜ ํฌ๊ธฐ๋ฅผ ๊ตฌ๋ถ„ํ•˜์ง€ ๋ชปํ•จ ์œ„ ๋ฌธ์ œ๋ฅผ .. 2023. 7. 10.
728x90
๋ฐ˜์‘ํ˜•