Skip to main content

Showing 1–7 of 7 results for author: Jeune, P L

Searching in archive cs. Search in all archives.
.
  1. arXiv:2505.11365  [pdf, other

    cs.CY cs.AI cs.CL cs.CR

    Phare: A Safety Probe for Large Language Models

    Authors: Pierre Le Jeune, Benoît Malézieux, Weixuan Xiao, Matteo Dora

    Abstract: Ensuring the safety of large language models (LLMs) is critical for responsible deployment, yet existing evaluations often prioritize performance over identifying failure modes. We introduce Phare, a multilingual diagnostic framework to probe and evaluate LLM behavior across three critical dimensions: hallucination and reliability, social biases, and harmful content generation. Our evaluation of 1… ▽ More

    Submitted 26 May, 2025; v1 submitted 16 May, 2025; originally announced May 2025.

  2. arXiv:2504.10277  [pdf

    cs.CY cs.AI cs.CL cs.CR

    RealHarm: A Collection of Real-World Language Model Application Failures

    Authors: Pierre Le Jeune, Jiaen Liu, Luca Rossi, Matteo Dora

    Abstract: Language model deployments in consumer-facing applications introduce numerous risks. While existing research on harms and hazards of such applications follows top-down approaches derived from regulatory frameworks and theoretical analyses, empirical evidence of real-world failure modes remains underexplored. In this work, we introduce RealHarm, a dataset of annotated problematic interactions with… ▽ More

    Submitted 14 April, 2025; originally announced April 2025.

  3. arXiv:2310.10433  [pdf, other

    cs.CV cs.LG

    Object Detection in Aerial Images in Scarce Data Regimes

    Authors: Pierre Le Jeune

    Abstract: Most contributions on Few-Shot Object Detection (FSOD) evaluate their methods on natural images only, yet the transferability of the announced performance is not guaranteed for applications on other kinds of images. We demonstrate this with an in-depth analysis of existing FSOD methods on aerial images and observed a large performance gap compared to natural images. Small objects, more numerous in… ▽ More

    Submitted 16 October, 2023; originally announced October 2023.

    Comments: PhD Thesis. Work conducted at L2TI (USPN) and COSE

  4. arXiv:2307.09562  [pdf, other

    cs.CV

    Rethinking Intersection Over Union for Small Object Detection in Few-Shot Regime

    Authors: Pierre Le Jeune, Anissa Mokraoui

    Abstract: In Few-Shot Object Detection (FSOD), detecting small objects is extremely difficult. The limited supervision cripples the localization capabilities of the models and a few pixels shift can dramatically reduce the Intersection over Union (IoU) between the ground truth and predicted boxes for small objects. To this end, we propose Scale-adaptive Intersection over Union (SIoU), a novel box similarity… ▽ More

    Submitted 17 July, 2023; originally announced July 2023.

  5. arXiv:2210.13923  [pdf, other

    cs.CV

    A Comparative Attention Framework for Better Few-Shot Object Detection on Aerial Images

    Authors: Pierre Le Jeune, Anissa Mokraoui

    Abstract: Few-Shot Object Detection (FSOD) methods are mainly designed and evaluated on natural image datasets such as Pascal VOC and MS COCO. However, it is not clear whether the best methods for natural images are also the best for aerial images. Furthermore, direct comparison of performance between FSOD methods is difficult due to the wide variety of detection frameworks and training strategies. Therefor… ▽ More

    Submitted 25 October, 2022; originally announced October 2022.

  6. arXiv:2201.02052  [pdf, other

    cs.CV

    A Unified Framework for Attention-Based Few-Shot Object Detection

    Authors: Pierre Le Jeune, Anissa Mokraoui

    Abstract: Few-Shot Object Detection (FSOD) is a rapidly growing field in computer vision. It consists in finding all occurrences of a given set of classes with only a few annotated examples for each class. Numerous methods have been proposed to address this challenge and most of them are based on attention mechanisms. However, the great variety of classic object detection frameworks and training strategies… ▽ More

    Submitted 6 January, 2022; originally announced January 2022.

  7. arXiv:2109.13027  [pdf, other

    cs.CV

    Experience feedback using Representation Learning for Few-Shot Object Detection on Aerial Images

    Authors: Pierre Le Jeune, Mustapha Lebbah, Anissa Mokraoui, Hanene Azzag

    Abstract: This paper proposes a few-shot method based on Faster R-CNN and representation learning for object detection in aerial images. The two classification branches of Faster R-CNN are replaced by prototypical networks for online adaptation to new classes. These networks produce embeddings vectors for each generated box, which are then compared with class prototypes. The distance between an embedding an… ▽ More

    Submitted 27 September, 2021; originally announced September 2021.

    Comments: 11 pages, 2 figures, accepted at ICMLA 2021 as a short paper