projet-long/slides.md
2023-06-25 20:03:29 +02:00

18 KiB

theme class coverBackgroundUrl coverBackgroundSource coverBackgroundSourceUrl coverDate themeConfig title
academic text-white https://plus.unsplash.com/premium_photo-1673553304257-018c85e606f8?ixlib=rb-4.0.3&ixid=MnwxMjA3fDB8MHxwaG90by1wYWdlfHx8fGVufDB8fHx8 unplash https://unsplash.com/photos/g4I556WCJT0
paginationX paginationY paginationPagesDisabled
r t
1
Projet Long

Sphere detection and multimedia applications

2023-03-09 Laurent Fainsin, Pierre-Eliot Jourdan, Raphaëlle Monville-Letu, Jade Neav

Contents

  • Types of spheres
  • Automatic sphere detection
  • Lighting intensity estimation
  • Lighting direction estimation
Architecture
Cinema
3D Reconstruction

pexels


class: text-white custombg

Types of spheres


class: text-white custombg2

Chrome sphere

CaveAcademy


Acquisition techniques

Louis du Mont


Realistic lighting

High Dynamic Range Imaging, Paul Debevec


class: text-white custombg3

Shiny sphere

CaveAcademy


class: text-white custombg4

Matte sphere

CaveAcademy


Automatic sphere detection

  • Model
  • Datasets
  • Results
  • Perspectives

Model

End-to-End Object Detection with Transformers, arXiv:2005.12872


Datasets (1/4)

Antoine Laurent


Datasets (2/4)

A Dataset of Multi-Illumination Images in the Wild


Datasets (3/4)

MS COCO compositing


Datasets (4/4)

Blender, PolyHaven

Results (1/8)


Results (2/8)


Results (3/8)


Results (4/8)


Results (5/8)


Results (6/8)


Results (7/8)


Results (8/8)


Perspectives

Poliigon.com


Lighting intensity estimation

  • Photometric Stereo
  • Lambert Law
  • Problem formulation
  • Algorithms
  • Generated images
  • Results
  • Perspectives

Photometric Stereo

  • Estimate the surface normals of an object
  • Shiny spheres \rightarrow direction of the lighting

Wikipedia


Lambert law

I(q) = \rho(Q) \times \vec{n}(Q) \cdot \vec{s}(Q)

  • \rho(Q) is the albedo

  • \vec{n}(Q) is the normal vector

  • \vec{s}(Q) = \phi \times \vec{s_0}(Q) is the lighting direction

LaserFocusWorld


Problem formulation

N lightings, P pixels
\rightarrow I = M \times S \times D_{\phi}

  • I \in \mathbb{R}^{P \times N} \rightarrow gray scale levels \rightarrow known from image pixels

  • M \in \mathbb{R}^{P \times 3} \rightarrow the albedo and the normals \rightarrow unknown

  • S \in \mathbb{R}^{3 \times N} \rightarrow direction of lightings \rightarrow known from shiny spheres

  • D_{phi} = diag(\phi_1,...,\phi_{N}) \in \mathbb{R}^{ N \times N} \rightarrow intensities of lightings \rightarrow to be determined


Algorithm 1

Intensities : [\phi_1,...,\phi_{N}]

New values : \phi_j \plusmn \delta, \ j \in [1,..,N]

Estimation of the matrix M

Mean-squared error : \underset{\phi_i}{\min} || I - M S D_{\phi} ||_2^2

Update the value of \phi_j

Repeat previous steps


Algorithm 2

Algorithm 1 \rightarrow too long

I = M S D_{\phi} \iff M = I(S D_{\phi})^\dagger = I (S D_{\phi})^T [(S D_{\phi})(S D_{\phi})^T]^{-1}

Lambert law :

$$ \begin{align*} I &= I (S D_{\phi})^T [(S D_{\phi})(S D_{\phi})^T]^{-1} S D_{\phi} \ &= I D_{\phi} S^T S^{-T} D_{\phi}^{-2} S^{-1} S D_{\phi} \end{align*}

New residual :

\underset{\phi_i}{\min} || I - I D_{\phi} S^T S^{-T} D_{\phi}^{-2} S^{-1} S D_{\phi} ||_2^2

Generated images


Results (1/2)


Results (2/2)


Real images


Results


Perspectives

3D reconstruction


Lighting direction estimation

  • Estimation of lighting vector
  • Neural Network
  • Real data
  • Generated data
  • Results
  • Perspectives

Estimation of lighting vector

flowchart LR
  id1[Bounding box of the sphere]
  id2[Deduce the normals]
  id3[Resolution of I = s * n]

  id1 --> id2
  id2 --> id3

Neural Network


ResNet-50


Real data : creation of mask


Generated data with blender

Simulated matte spheres
Generated data with different lightings

Results


Perspectives

  • Create more data to prevent overfitting
  • Diversify the types of data lighting (more than 8 directions)
  • Transform the model into something more general:
    \rightarrow from {image of sphere, vector lighting} to {image of objects, vector lighting}