All posts
Papers · paper Nagaonkar, Garg, Raj, Choithani, Trivedi

Search over the Visual World: persistent visual memory, layered indexes, and source-grounded evidence

VideoDB technical report (arXiv 2608.08075). Analyzer-defined scenes, persistent visual memory, capability-declared indexes, source-grounded evidence, and a 9,834-query retrieval evaluation against a commercial video-native engine.

Sankalp Nagaonkar · Rohit Garg · Ankit Raj · Ashish Choithani · Ashutosh Trivedi
VideoDB · {sankalp, rohit, ankit, ashish, ashu}@videodb.io
Technical Report · July 2026

Read on arXiv · Download PDF · Benchmark code

Search over the visual world cannot be reduced to ranking video files. This report develops the infrastructure it does require: analyzer-defined scenes, persistent visual memory, capability-declared indexes, and evidence that stays playable at the source. It then evaluates it against a commercial video-native retrieval engine over 9,834 natural-language queries drawn from four public datasets. The paper is published on arXiv (2608.08075).

Trouble viewing it here? Open the PDF in a new tab. Benchmark configurations and reproduction instructions are on GitHub, and the open-source Deep Search implementation of the stateful retrieval loop is public too.

Cite this work

S Nagaonkar, Rohit Garg, Ankit Raj, Ashish Choithani, Ashutosh Trivedi, "Search over the Visual World: persistent visual memory, layered indexes, and source-grounded evidence", arXiv preprint arXiv:2608.08075, August 2026.

@article{nagaonkar2026searchvisualworld,
  author = {S Nagaonkar and Rohit Garg and Ankit Raj and Ashish Choithani and Ashutosh Trivedi},
  title = {Search over the Visual World: persistent visual memory, layered indexes, and source-grounded evidence},
  journal = {arXiv preprint arXiv:2608.08075},
  year = {2026},
  month = {aug},
  note = {https://arxiv.org/abs/2608.08075},
}
Machine

https://videodb.io/blog/search-over-the-visual-world.mdOpen the file