Picture for Piotr Koniusz

Piotr Koniusz

INRIA Grenoble Rhône-Alpes / LJK Laboratoire Jean Kuntzmann

MATHGLANCE: Multimodal Large Language Models Do Not Know Where to Look in Mathematical Diagrams

Add code
Mar 26, 2025
Viaarxiv icon

Graph Self-Supervised Learning with Learnable Structural and Positional Encodings

Add code
Feb 22, 2025
Figure 1 for Graph Self-Supervised Learning with Learnable Structural and Positional Encodings
Figure 2 for Graph Self-Supervised Learning with Learnable Structural and Positional Encodings
Figure 3 for Graph Self-Supervised Learning with Learnable Structural and Positional Encodings
Figure 4 for Graph Self-Supervised Learning with Learnable Structural and Positional Encodings
Viaarxiv icon

Open Eyes, Then Reason: Fine-grained Visual Mathematical Understanding in MLLMs

Add code
Jan 11, 2025
Viaarxiv icon

When Spatial meets Temporal in Action Recognition

Add code
Nov 22, 2024
Figure 1 for When Spatial meets Temporal in Action Recognition
Figure 2 for When Spatial meets Temporal in Action Recognition
Figure 3 for When Spatial meets Temporal in Action Recognition
Figure 4 for When Spatial meets Temporal in Action Recognition
Viaarxiv icon

Inductive Graph Few-shot Class Incremental Learning

Add code
Nov 11, 2024
Viaarxiv icon

Less is More: Extreme Gradient Boost Rank-1 Adaption for Efficient Finetuning of LLMs

Add code
Oct 25, 2024
Viaarxiv icon

LEGO: Learnable Expansion of Graph Operators for Multi-Modal Feature Fusion

Add code
Oct 02, 2024
Viaarxiv icon

OpenKD: Opening Prompt Diversity for Zero- and Few-shot Keypoint Detection

Add code
Sep 30, 2024
Figure 1 for OpenKD: Opening Prompt Diversity for Zero- and Few-shot Keypoint Detection
Figure 2 for OpenKD: Opening Prompt Diversity for Zero- and Few-shot Keypoint Detection
Figure 3 for OpenKD: Opening Prompt Diversity for Zero- and Few-shot Keypoint Detection
Figure 4 for OpenKD: Opening Prompt Diversity for Zero- and Few-shot Keypoint Detection
Viaarxiv icon

PACE: marrying generalization in PArameter-efficient fine-tuning with Consistency rEgularization

Add code
Sep 25, 2024
Viaarxiv icon

PseudoNeg-MAE: Self-Supervised Point Cloud Learning using Conditional Pseudo-Negative Embeddings

Add code
Sep 24, 2024
Viaarxiv icon