🧠 Shared Tasks @ WSLP 2025 – Sign Language
⚠️ Interested in participating?
Join our Discord server to stay updated and connect with other participants!
🧾 Call for Shared Tasks
WSLP 2025: Workshop on Sign Language Translation
- 📍 Co-located with IJCNLP-AACL 2025
- 📅 Date: TBD
- 📍 Victor Menezes Convention Centre (VMCC), IIT Bombay, Mumbai, India
WSLP 2025 invites participants to three Shared Tasks addressing key challenges in Indian Sign Language (ISL) processing.
🔍 Shared Tasks Overview
| Tasks | Focus | Public Dataset | Task dataset | Codabench |
|---|---|---|---|---|
| ISL to English Translation | Sentence-level translation from ISL videos/pose to English | iSign | Translation Dataset | Translation Codabench |
| Word/Gloss Recognition | Isolated sign recognition at the word level | CISLR | word Recognition Dataset | Word Recognition Codabench |
| Word Presence Prediction | Detecting presence of a word in a signed sentence | Word Presence Dataset | Word Presence Prediction Dataset | Word Presence Prediction Codabench |
📋 Shared Tasks Details
🌐 Task 1: ISL to English Translation
Goal: Translate sentence-level Indian Sign Language (ISL) videos/poses into English text.
- Challenges: Visual-linguistic grounding, grammar, gesture ambiguity
- Use Cases: Sign-enabled chatbots, video interpreters, accessible interfaces
- Dataset: Public Dataset iSign (118,000 video-sentence pairs) , Task validation and Test dataset Translation Dataset
- Metrics: BLEU, ROUGE, chrF
- Input ➝ Output: ISL video/pose ➝ English sentence
English Translation: “The ban would mean she can’t compete in any national or other domestic events”
✋ Task 2: Word/Gloss Recognition
Goal: Recognize isolated ISL signs (words or glosses) from short video clips.
- Challenges: Sign variability, subtle motion, similar gestures
- Use Cases: Dictionary building, lookup tools, annotation
- Dataset: Dataset word Recognition Dataset
- Metrics: Accuracy, Top-K Accuracy
- Input ➝ Output: Video clip ➝ Word label
Label: “National”
🔍 Task 3: Word Presence Prediction
Goal: Predict if a given word is present in a full ISL sentence video.
- Challenges: Sign spotting, context alignment
- Use Cases: Query-based video retrieval, sign search
- Dataset: Word Presence Dataset
- Metrics: Accuracy, Precision, Recall, F1
- Input ➝ Output: (Video, Word) ➝ Present / Not Present
Query Word: “National”
Sentence Video: “The ban would mean she can’t compete in any national or other domestic events” contains: “National”
🗓 Key Dates
| Event | Date |
|---|---|
| 🟢 Start Date | August 15, 2025 |
| 📚 Training Phase | August 15 – October 5, 2025 |
| 🧪 Testing Phase | October 5 – October 15, 2025 |
| 📄 Paper Submission Deadline | October 25, 2025 |
| 📬 Notification of Acceptance | November 3, 2025 |
| 📸 Camera-ready Papers Due | November 11, 2025 |
| 📚 Proceedings Due | December 1, 2025 |
👥 Call for Participation
We invite researchers, students, and developers in computer vision, natural language processing, speech and gesture technology, or related fields to participate. Contribute to building inclusive tools for millions of ISL users.
Contact Organizers:
- Abhinav Joshi: ajoshi@cse.iitk.ac.in
- Sanjeet Singh: sanjeet@cse.iitk.ac.in
📦 Submission Guidelines
- Platform: Codabench
- Team Size: Max 4 members (TBD)
- Format: To be released with dataset
- Requirements: Output on test set + short documentation
- Paper Submission: Full (8 pages) or short (4 pages) papers, following ACL Style. Double-blind review. Accepted papers get +1 page for revisions.
✅ Get ready to build impactful AI tools for the Deaf community.