# ZipDepth Distills a Foundation Depth Model Into 6.1 Million Parameters That Run on Phones

Trained on 14.1 million images across 17 domains via distillation from Depth Anything v2-Large, the network exports to TensorRT, CoreML, and ONNX without manual graph rewriting.

- Published: 2026-07-11T05:46:26.030Z
- Canonical: https://polylog.news/ai/2026-07-11/zipdepth-distills-a-foundation-depth-model-into-6-1-million
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news), [ZipDepth (project page)](https://zipdepth.github.io/)

A team from the University of Bologna released ZipDepth, a compact monocular depth network aimed at running foundation-quality depth estimation on embedded and mobile hardware. It pairs a reparameterizable encoder-decoder with large-scale k…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-07-11/zipdepth-distills-a-foundation-depth-model-into-6-1-million (subscription information: https://polylog.news/pricing).