2nd Call-for-Participation: ImageCLEF 2025 Multimedia Retrieval in CLEF Lab

[Apologies for multiple postings]

ImageCLEF 2025
Multimedia Retrieval in CLEF
http://www.imageclef.org/2025/

*** CALL FOR PARTICIPATION ***

ImageCLEF 2025 is an evaluation campaign that is being organized as part =
of the CLEF (Conference and Labs of the Evaluation Forum) labs. The campa=
ign offers several research tasks that welcome participation from teams a=
round the world.

The results of the campaign appear in the working notes proceedings, publ=
ished by CEUR Workshop Proceedings (CEUR-WS.org) and are presented in the=
 CLEF conference. Selected contributions among the participants, will be =
invited for publication in the following year in the Springer Lecture Not=
es in Computer Science (LNCS) together with the annual lab overviews.

Target communities involve (but are not limited to): information retrieva=
l (text, vision, audio, multimedia, social media, sensor data, etc.), mac=
hine learning, deep learning, data mining, natural language processing, i=
mage and video processing, computer vision, with special attention to the=
 challenges of multi-modality, multi-linguality, and interactive search.


*** 2025 TASKS ***
ImageCLEFmedical Automatic Image Captioning
ImageCLEFmedical Synthetic Medical Images Created via GANs
ImageCLEFmedical Visual Question Answering
ImageCLEFmedical Multimodal And Generative TelemedICine (MAGIC)
Image Retrieval/Generation for Arguments
ImageCLEFtoPicto
ImageCLEF Multimodal Reasoning

#ImageCLEFmedical Automatic Image Captioning (9th edition)
https://www.imageclef.org/2025/medical/caption
Interpreting and summarizing the insights gained from medical images such=
 as radiology output is a time-consuming task that involves highly traine=
d experts and often represents a bottleneck in clinical diagnosis pipelin=
es.The Automatic Image Captioning task is split into 2 subtasks: Concept =
Detection Task, based on identifying the presence and location of relevan=
t concepts in a large corpus of medical images and the Caption Prediction=
 Task, where participating systems are tasked with composing coherent cap=
tions for the entirety of an image

Organizers: Hendrik Damm, Johannes R=C3=BCckert, Christoph M. Friedrich, =
Louise Bloch, Raphael Br=C3=BCngel, Ahmad Idrissi-Yaghir, Benjamin Bracke=
 (University of Applied Sciences and Arts Dortmund, Germany), Asma Ben Ab=
acha (Microsoft, USA), Alba Garc=C3=ADa Seco de Herrera (University of Es=
sex, UK), Henning M=C3=BCller (University of Applied Sciences Western Swi=
tzerland, Sierre, Switzerland), Henning Sch=C3=A4fer, Tabea M. G. Pakull =
(Institute for Transfusion Medicine, University Hospital Essen, Germany),=
 Cynthia S. Schmidt, Obioma Pelka (Institute for Artificial Intelligence =
in Medicine, Germany)


#ImageCLEFmedical Synthetic Medical Images Created via GANs (3rd edition)=

https://www.imageclef.org/2025/medical/gan
The task aims to further investigate the hypothesis that generative model=
s generate synthetic medical images that retain "fingerprints" from the r=
eal images used during their training. These fingerprints raise important=
 security and privacy concerns, particularly in the context of personal m=
edical image data being used to create artificial images for various real=
-life applications. In the first subtask, participants will analyze synth=
etic biomedical images to determine whether specific real images were use=
d in the training process of generative models. In the second subtask, pa=
rticipants will link each synthetic biomedical image to the specific subs=
et of real data used during its generation. The goal is to identify the p=
articular dataset of real images that contributed to the training of the =
generative model responsible for creating each synthetic image.

Organizers: Alexandra Andrei, Liviu-Daniel =C8=98tefan, Mihai Gabriel Con=
stantin, Mihai Dogariu, Bogdan Ionescu (National University of Science an=
d Technology POLITEHNICA Bucharest, Romania), Ahmedkhan Radzhabov, Yuri P=
rokopchuk (National Academy of Science of Belarus, Minsk, Belarus), Vassi=
li Kovalev (Belarusian Academy of Sciences, Minsk, Belarus), Henning M=C3=
=BCller (University of Applied Sciences Western Switzerland, Sierre, Swit=
zerland)


#ImageCLEFmedical Visual Question Answering (3rd edition)
https://www.imageclef.org/2025/medical/vqa
This year, the challenge looks at the integration of Visual Question Answ=
ering (VQA) with synthetic gastrointestinal (GI) data, aiming to enhance =
diagnostic accuracy and learning algorithms. The challenge includes devel=
oping algorithms that can interpret and answer questions based on synthet=
ic GI images, creating advanced synthetic images that mimic accurate diag=
nostic visuals in detail and variability, and evaluating the effectivenes=
s of VQA techniques with both synthetic and real GI data.
The 1st subtask asks participants to build algorithms that can accurately=
 interpret and respond to questions pertaining to gastrointestinal (GI) i=
mages. This involves understanding the context and details within the ima=
ges and providing precise answers that would assist in medical diagnostic=
s, while the 2nd subtask focuses on the generation of synthetic GI images=
 that are highly detailed and variable enough to closely resemble real me=
dical images.

Organizers: Steven A. Hicks, Sushant Gautam, Michael A. Riegler, Vajira T=
hambawita, P=C3=A5l Halvorsen (SimulaMet, Norway)

#ImageCLEFmedical Multimodal And Generative TelemedICine (MEDIQA-MAGIC) (=
3rd edition)
https://www.imageclef.org/2025/medical/mediqa
The task extends on the previous year=E2=80=99s dataset and challenge bas=
ed on multimodal dermatology response generation. Participants will be gi=
ven a clinical narrative context along with accompanying images. The task=
 is divided into two relevant sub-parts: (i) segmentation of dermatologic=
al problem regions, and (ii) providing answers to closed-ended questions =
(participants will be given a dermatological query, its accompanying imag=
es, as well as a closed-question with accompanying choices =E2=80=93 the =
task is to select the correct answer to each question)

Organizers: Asma Ben Abacha, Wen-wai Yim, Noel Codella (Microsoft), Rober=
to Andres Novoa (Stanford University), Josep Malvehy (Hospital Clinic of =
Barcelona)

#Image Retrieval/Generation for Arguments  (4th edition)
https://www.imageclef.org/2025/argument-images
Given a set of arguments, the task is to return for each argument several=
 images that help convey the argument. A suitable image could depict the =
argument or show a generalization or specialization. Participants can opt=
ionally add a short caption that explains the meaning of the image. Image=
s can be either retrieved from the focused crawl or generated using an im=
age generator.

Organizers: Maximilian Heinrich, Johannes Kiesel, Benno Stein (Bauhaus-Un=
iversit=C3=A4t Weimar), Moritz Wolter (Leipzig University), Martin Pottha=
st (University of Kassel, hessian.AI, scads.AI)

#ImageCLEFtoPicto (3rd edition)
https://www.imageclef.org/2025/topicto
The goal of ToPicto is to bring together linguists, computer scientists, =
and translators to develop new translation methods to translate either sp=
eech or text into a corresponding sequence of pictograms. The task refers=
 to the relationship between text and related pictograms and is composed =
of 2 subtasks: the Text-to-Picto task, which focuses on the automatic gen=
eration of a corresponding sequence of pictogram terms and the Speech-to-=
Picto task, which focuses on directly translating speech to pictogram ter=
ms.

Organizers: Diandra Fabre, C=C3=A9cile Macaire, Benjamin Lecouteux, Didie=
r Schwab (Universit=C3=A9 Grenoble Alpes, LIG, France)

#ImageCLEF Multimodal Reasoning (new)
https://www.imageclef.org/2025/multimodalreasoning
MultimodalReason is a new task focusing on Multilingual Visual Question A=
nswering (VQA). The formulation of the task is the following: Given an im=
age of a question with 3-5 possible answers, participants must identify t=
he single correct answer.The task is split into many subtasks, each handl=
ing a different language (English, Bulgarian, Arabic, Serbian, Italian, H=
ungarian, Croatian, Urdu, Kazakh, Spanish, with a few more on the way). T=
he task's goal is to assess modern LLMs' reasoning capabilities on comple=
x inputs, presented in different languages, across various subjects.

Organizers: Dimitar Dimitrov, Ivan Koychev (Sofia University "St. Kliment=
 Ohridski", Bulgaria), Rocktim Jyoti Das, Zhuohan Xie, Preslav Nakov (Moh=
amed bin Zayed University of Artificial Intelligence (MBZUAI), Abu Dhabi,=
 UAE)



*** IMPORTANT DATES ***
(may vary depending on the task)
- Run submission: May 10, 2025
- Working notes submission: May 30, 2025
- CLEF 2025 conference: September 9-12, 2025, Madrid, Spain


*** REGISTRATION ***
Follow the instructions here https://www.imageclef.org/2025


*** OVERALL COORDINATION ***
Bogdan Ionescu, Politehnica University of Bucharest, Romania
Henning M=C3=BCller, HES-SO, Sierre, Switzerland
Cristian Stanciu, Politehnica University of Bucharest, Romania



On behalf of the organizers,

Cristian Stanciu
https://www.aimultimedialab.ro/

########################################################################

To unsubscribe from the IR list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=3DIR&A=3D1

This message was issued to members of www.jiscmail.ac.uk/IR, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/