Introducing the Visual AI Track at The Fifth Elephant 2025

Hi everyone! I’m thrilled to announce that I’ll be curating the Visual AI Track at The Fifth Elephant 2025 conference.

When CLIP first came out, it looked almost like a party trick: show it an image and it could auto-complete a caption like “a photo of…” with uncanny accuracy. I remember thinking, that’s neat — but I had no idea it would end up powering everything from Stable Diffusion to the explicit content filter in one of our production systems.

The wild part? It was right there all along — we just hadn’t recognized its full potential yet.

That’s why I’m excited to curate this track. Sometimes, the most impactful ideas don’t look big at first. They grow through use, feedback, and iteration. This track is a space for sharing those small sparks, overlooked insights, and hard-earned lessons from building with Visual AI.

A bit about me

My name is Sidharth and I lead multiple data and AI teams at a large European media company. We work with vision and language models for various use-cases like explicit content detection, creation of marketing and video assets and more. I’ve been an active member of the Fifth Elephant community — I presented the StableDiffusion paper, led an LLM fine-tuning/script-writing workshop, and learned alongside so many of you. I’m especially excited to bring this experience into curating a track that’s both technically rich and grounded in real-world practice.

What’s this track about — and who it’s for?

The Visual AI track is designed for researchers, engineers, developers, content creators, and technical leaders working with AI for visual applications. If you’re building or using systems that analyze, generate, manage, or moderate visual content — this is your tribe.
We’re focusing on real-world deployment stories, practical use-cases, and lessons learned — not just glossy demos. Whether you’re in entertainment, gaming, infrastructure, digital safety, or industrial AI, your insights can help shape how we build and use these technologies responsibly.

CFP Banner

Topics we’re looking for

Here’s a non-exhaustive list of themes we’d love to explore:

Text-to-video and generative visual content systems
Content moderation and digital safety using computer vision
Real-time video analytics in immersive media and gaming
Photogrammetry and 3D modeling for industrial applications
AI-generated virtual environments and digital twins
Deploying vision systems in the wild (autonomous vehicles, warehouses, etc.)
Building tools, frameworks, and platforms for Visual AI
Evaluation, benchmarking, and quality assurance for visual models
Responsible and ethical use of visual data and AI
Case studies with honest lessons from production deployments

Who should consider speaking

If you’ve:

Experimented with visual models in production,
Built scrappy but effective vision-based systems,
Faced the messiness of real-world data,
Integrated visual AI into an existing workflow or product,
…then we want to hear from you!

You don’t need a massive paper or a polished product. If you’ve got an insight, a story, or even a bug that taught you something valuable — that’s worth sharing.

The deadline for submissions is 31st May. I encourage everyone to submit your talk idea and promise to provide feedback and help you succeed as a speaker.

Submit your proposal →

Looking forward to hosting a diverse set of voices and experiences in the Visual AI space. Let’s build this together.

The Fifth Elephant 2025 Annual Conference CfP

{{ gettext('Draft') }}