PhD student in Applied Mathematics & Data Science at FGV EMAp, Rio de Janeiro 🇧🇷 — researching at the VisualDS Lab under Jorge Poco. I work at the intersection of computer vision, urban computing and visual analytics: turning raw video from cities into something people can actually explore and reason about.
- 🔬 Currently working on multi-camera pedestrian tracking, 3D pose reconstruction and visual analytics for urban video
- 🧰 I also run the engineering side of the lab: CI/CD, Docker/Portainer, GPU workstations, private networking
- 🇵🇪 Originally from Puno, Peru (CS @ UNSA) — 🇧🇷 based in Rio de Janeiro
- 🌐 Projects & writing at reqhiem.dev
- 📫 Reach me at joelperca.vc@gmail.com
- UrbanClipAtlas: A Visual Analytics Framework for Urban Video Analysis J. Perca, L. Sante, J. Heredia, J. Rulff, C. Silva, J. Poco Computer Graphics Forum (EuroVis 2026 Proceedings), vol. 45, no. 3
- UrbanClipAtlas — foundation-model-driven exploration of large urban video collections (FastAPI · Svelte · D3.js)
- StreetAware — multi-camera synchronized pedestrian tracking (YOLO · SAM · DuckDB)
- 3D multi-person reconstruction — tracking + pose lifting across calibrated camera rigs
