Ewa Kijak
YOU?
Author Swipe
View article: Fast, Secure, and High-Capacity Image Watermarking with Autoencoded Text Vectors
Fast, Secure, and High-Capacity Image Watermarking with Autoencoded Text Vectors Open
Most image watermarking systems focus on robustness, capacity, and imperceptibility while treating the embedded payload as meaningless bits. This bit-centric view imposes a hard ceiling on capacity and prevents watermarks from carrying use…
View article: Flyweight FLIM Networks for Salient Object Detection in Biomedical Images
Flyweight FLIM Networks for Salient Object Detection in Biomedical Images Open
Salient Object Detection (SOD) with deep learning often requires substantial computational resources and large annotated datasets, making it impractical for resource-constrained applications. Lightweight models address computational demand…
View article: Adapting Without Seeing: Text-Aided Domain Adaptation for Adapting CLIP-like Models to Novel Domains
Adapting Without Seeing: Text-Aided Domain Adaptation for Adapting CLIP-like Models to Novel Domains Open
International audience
View article: Reframing Image Difference Captioning with BLIP2IDC and Synthetic Augmentation
Reframing Image Difference Captioning with BLIP2IDC and Synthetic Augmentation Open
The rise of the generative models quality during the past years enabled the generation of edited variations of images at an important scale. To counter the harmful effects of such technology, the Image Difference Captioning (IDC) task aims…
View article: SWIFT: Semantic Watermarking for Image Forgery Thwarting
SWIFT: Semantic Watermarking for Image Forgery Thwarting Open
This paper proposes a novel approach towards image authentication and tampering detection by using watermarking as a communication channel for semantic information. We modify the HiDDeN deep-learning watermarking architecture to embed and …
View article: Beyond Internet Images: Evaluating Vision-Language Models for Domain Generalization on Synthetic-to-Real Industrial Datasets
Beyond Internet Images: Evaluating Vision-Language Models for Domain Generalization on Synthetic-to-Real Industrial Datasets Open
International audience
View article: Distinctive Image Captioning: Leveraging Ground Truth Captions in CLIP Guided Reinforcement Learning
Distinctive Image Captioning: Leveraging Ground Truth Captions in CLIP Guided Reinforcement Learning Open
Training image captioning models using teacher forcing results in very generic samples, whereas more distinctive captions can be very useful in retrieval applications or to produce alternative texts describing images for accessibility. Rei…
View article: Embedding Space Interpolation Beyond Mini-Batch, Beyond Pairs and Beyond Examples
Embedding Space Interpolation Beyond Mini-Batch, Beyond Pairs and Beyond Examples Open
Mixup refers to interpolation-based data augmentation, originally motivated as a way to go beyond empirical risk minimization (ERM). Its extensions mostly focus on the definition of interpolation and the space (input or feature) where it t…
View article: Learning on graphs and hierarchies
Learning on graphs and hierarchies Open
Hierarchies, as described in mathematical morphology, represent nested regions of interest that facilitate high-level analysis and provide mechanisms for coherent data organization. Represented as hierarchical trees, they have formalisms i…
View article: MAAIP: Multi-Agent Adversarial Interaction Priors for imitation from fighting demonstrations for physics-based characters
MAAIP: Multi-Agent Adversarial Interaction Priors for imitation from fighting demonstrations for physics-based characters Open
Simulating realistic interaction and motions for physics-based characters is of great interest for interactive applications, and automatic secondary character animation in the movie and video game industries. Recent works in reinforcement …
View article: Building Flyweight FLIM-based CNNs with Adaptive Decoding for Object Detection
Building Flyweight FLIM-based CNNs with Adaptive Decoding for Object Detection Open
State-of-the-art (SOTA) object detection methods have succeeded in several applications at the price of relying on heavyweight neural networks, which makes them inefficient and inviable for many applications with computational resource con…
View article: AIP: Adversarial Interaction Priors for Multi-Agent Physics-based Character Control
AIP: Adversarial Interaction Priors for Multi-Agent Physics-based Character Control Open
We address the problem of controlling and simulating interactions between multiple physics-based characters, using short unlabeled motion clips. We propose Adversarial Interaction Priors (AIP), a multi-agents generative adversarial imitati…
View article: Which Discriminator for Cooperative Text Generation?
Which Discriminator for Cooperative Text Generation? Open
Language models generate texts by successively predicting probability\ndistributions for next tokens given past ones. A growing field of interest\ntries to leverage external information in the decoding process so that the\ngenerated texts …
View article: Teach me how to Interpolate a Myriad of Embeddings
Teach me how to Interpolate a Myriad of Embeddings Open
Mixup refers to interpolation-based data augmentation, originally motivated as a way to go beyond empirical risk minimization (ERM). Yet, its extensions focus on the definition of interpolation and the space where it takes place, while the…
View article: HDR-LFNet: Inverse Tone Mapping using Fusion Network
HDR-LFNet: Inverse Tone Mapping using Fusion Network Open
To capture the real-world luminance values, High Dynamic Range (HDR) image processing has been developed. HDR images have a richer content than the widely-used Standard Dynamic Range (SDR) images, and are used in a number of situations, e.…
View article: Generative Cooperative Networks for Natural Language Generation
Generative Cooperative Networks for Natural Language Generation Open
Generative Adversarial Networks (GANs) have known a tremendous success for many continuous generation tasks, especially in the field of image generation. However, for discrete outputs such as language, optimizing GANs remains an open probl…
View article: PPL-MCTS: Constrained Textual Generation Through Discriminator-Guided MCTS Decoding
PPL-MCTS: Constrained Textual Generation Through Discriminator-Guided MCTS Decoding Open
International audience
View article: PPL-MCTS: Constrained Textual Generation Through Discriminator-Guided Decoding
PPL-MCTS: Constrained Textual Generation Through Discriminator-Guided Decoding Open
International audience
View article: Generating artificial texts as substitution or complement of training\n data
Generating artificial texts as substitution or complement of training\n data Open
The quality of artificially generated texts has considerably improved with\nthe advent of transformers. The question of using these models to generate\nlearning data for supervised learning tasks naturally arises. In this article,\nthis qu…
View article: Generating artificial texts as substitution or complement of training data
Generating artificial texts as substitution or complement of training data Open
The quality of artificially generated texts has considerably improved with the advent of transformers. The question of using these models to generate learning data for supervised learning tasks naturally arises. In this article, this quest…
View article: AlignMix: Improving representations by interpolating aligned features
AlignMix: Improving representations by interpolating aligned features Open
Mixup is a powerful data augmentation method that interpolates between two or more examples in the input or feature space and between the corresponding target labels. Many recent mixup methods focus on cutting and pasting two or more objec…
View article: PPL-MCTS: Constrained Textual Generation Through Discriminator-Guided MCTS Decoding
PPL-MCTS: Constrained Textual Generation Through Discriminator-Guided MCTS Decoding Open
Large language models (LM) based on Transformers allow to generate plausible long texts. In this paper, we explore how this generation can be further controlled at decoding time to satisfy certain constraints (e.g. being non-toxic, conveyi…
View article: PPL-MCTS: Constrained Textual Generation Through Discriminator-Guided MCTS Decoding
PPL-MCTS: Constrained Textual Generation Through Discriminator-Guided MCTS Decoding Open
Large language models (LM) based on Transformers allow to generate plausible long texts. In this paper, we explore how this generation can be further controlled at decoding time to satisfy certain constraints (e.g. being non-toxic, conveyi…
View article: It Takes Two to Tango: Mixup for Deep Metric Learning
It Takes Two to Tango: Mixup for Deep Metric Learning Open
Metric learning involves learning a discriminative representation such that embeddings of similar classes are encouraged to be close, while embeddings of dissimilar classes are pushed far apart. State-of-the-art methods focus mostly on sop…
View article: AlignMixup: Improving Representations By Interpolating Aligned Features
AlignMixup: Improving Representations By Interpolating Aligned Features Open
Mixup is a powerful data augmentation method that interpolates between two or more examples in the input or feature space and between the corresponding target labels. Many recent mixup methods focus on cutting and pasting two or more objec…
View article: Detecting Human-Object Interaction with Mixed Supervision
Detecting Human-Object Interaction with Mixed Supervision Open
Human object interaction (HOI) detection is an important task in image understanding and reasoning. It is in a form of HOI triplet , requiring bounding boxes for human and object, and action between them for the task completion. In other w…
View article: Combining convolutional side-outputs for road image segmentation
Combining convolutional side-outputs for road image segmentation Open
International audience
View article: Fusion-based multimodal detection of hoaxes in social networks
Fusion-based multimodal detection of hoaxes in social networks Open
International audience
View article: Context-Aware Forgery Localization in Social-Media Images: A Feature-Based Approach Evaluation
Context-Aware Forgery Localization in Social-Media Images: A Feature-Based Approach Evaluation Open
International audience