Art2Contour: Salient Contour Detection in Artworks Using Generative Adversarial Networks

Sindel, Aline; Maier, Andreas; Christlein, Vincent

doi:10.1109/icip40778.2020.9191117

Cited by 5 publications

(5 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Analogously, we also extract the same regions from the upscaled feature maps of the global network, which is the direct output of the global network before applying spatial softmax. All regions are 20). With the channel fusion, the region network gets the global location of the landmark as prior information, which supports the refinement task.…”

Section: Regional Facial Landmarks Refinementmentioning

confidence: 86%

“…Another possibility is to focus only on the facial contours for the comparison. For this task, we generate contour drawings for the three examples using the conditional generative adversarial network Art2Contour [20] and apply the same transformations that we have predicted using the facial landmarks to warp the contour images correspondingly (see (h)-(j) in Fig. 8).…”

Section: Facial Image and Contour Comparisonmentioning

confidence: 99%

See 1 more Smart Citation

ArtFacePoints: High-resolution Facial Landmark Detection in Paintings and Prints

Sindel¹,

Maier²,

Christlein³

2022

Preprint

View full text Add to dashboard Cite

Facial landmark detection plays an important role for the similarity analysis in artworks to compare portraits of the same or similar artists. With facial landmarks, portraits of different genres, such as paintings and prints, can be automatically aligned using control-pointbased image registration. We propose a deep-learning-based method for facial landmark detection in high-resolution images of paintings and prints. It divides the task into a global network for coarse landmark prediction and multiple region networks for precise landmark refinement in regions of the eyes, nose, and mouth that are automatically determined based on the predicted global landmark coordinates. We created a synthetically augmented facial landmark art dataset including artistic style transfer and geometric landmark shifts. Our method demonstrates an accurate detection of the inner facial landmarks for our high-resolution dataset of artworks while being comparable for a public low-resolution artwork dataset in comparison to competing methods.

show abstract

Section: Regional Facial Landmarks Refinementmentioning

confidence: 86%

Section: Facial Image and Contour Comparisonmentioning

confidence: 99%

ArtFacePoints: High-resolution Facial Landmark Detection in Paintings and Prints

Sindel¹,

Maier²,

Christlein³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…The adversarial loss is combined with a binary cross-entropy content loss for which the set of ground truth contour images is linearly merged into a single ground truth image. Art2Contour [19] utilizes a conditional GAN with a ResNet-based generator network for salient contour detection in prints and paintings. Art2Contour is trained with a combined loss of the cGAN loss and a task loss consisting of multiple regression terms, which separately treat the single ground truth images.…”

Section: Contour Detection Using Generative Adversarial Networkmentioning

confidence: 99%

“…Our chain line segmentation network is a conditional generative adversarial network (cGAN) [20] consisting of a generator and discriminator network. Our generator network is the ResNet-based [21] encoder-decoder architecture that was introduced for style transfer [22], having ResNet blocks in the bottleneck, and in contrast to UNet [7], it does not have skip connections between the encoder and decoder [19,23]. As the discriminator network, we used a regular global GAN that has been shown to be effective for contour detection [19,23].…”

Section: Chain Line Segmentation Network Architecturementioning

confidence: 99%

“…Our method was implemented using the PyTorch framework, and the end-to-end training and inference both ran completely on the GPU. The generator network (9 ResNet blocks) and the discriminator network were trained from scratch for 100 epochs with early stopping by using the Adam optimizer, a learning rate of η = 0.0002 with linear decay to 0 starting at Epoch 50, momentum (0.5, 0.999), a batch size of 2, λ 0 = 1000 [19], λ BCE = 0.5, λ DICE = 0.5, λ DSAC = 0.5, and λ MLE = 0.5. For DSAC, m = 64 hypotheses are sampled based on k p = 500 points from each bounding box per patch or k p = 1300 points from each bounding box per image.…”

Section: Implementation Detailsmentioning

confidence: 99%

See 1 more Smart Citation

ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical Prints

et al. 2021

Self Cite

View full text Add to dashboard Cite

The paper structure of historical prints is sort of a unique fingerprint. Paper with the same origin shows similar chain line distances. As the manual measurement of chain line distances is time consuming, the automatic detection of chain lines is beneficial. We propose an end-to-end trainable deep learning method for segmentation and parameterization of chain lines in transmitted light images of German prints from the 16th Century. We trained a conditional generative adversarial network with a multitask loss for line segmentation and line parameterization. We formulated a fully differentiable pipeline for line coordinates’ estimation that consists of line segmentation, horizontal line alignment, and 2D Fourier filtering of line segments, line region proposals, and differentiable line fitting. We created a dataset of high-resolution transmitted light images of historical prints with manual line coordinate annotations. Our method shows superior qualitative and quantitative chain line detection results with high accuracy and reliability on our historical dataset in comparison to competing methods. Further, we demonstrated that our method achieves a low error of less than 0.7 mm in comparison to manually measured chain line distances.

show abstract

ArtFacePoints: High-Resolution Facial Landmark Detection in Paintings and Prints

Sindel

Maier

Christlein

2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Art2Contour: Salient Contour Detection in Artworks Using Generative Adversarial Networks

Cited by 5 publications

References 15 publications

ArtFacePoints: High-resolution Facial Landmark Detection in Paintings and Prints

ArtFacePoints: High-resolution Facial Landmark Detection in Paintings and Prints

ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical Prints

ArtFacePoints: High-Resolution Facial Landmark Detection in Paintings and Prints

Contact Info

Product

Resources

About