FocsleFo’c’sle
Fo’c’sle

ZoeDepth-Mobile

7%
by Fo’c’sle

Metric depth in real units. Calibrated for mobile-tier NPUs.

DepthFocsle-ResearchINT8FP16metric-depthindoor
41K downloads 1.8K deploymentsUpdated Feb 9, 2028
Headline:28.4ms · Qualcomm QCS8550 · INT8
OverviewBenchmarks5Sim ResultsDeploy5FilesDiscussion23

About this model

Metric depth in real units. Calibrated for mobile-tier NPUs.

Authored by focsle. Curated into the Fo’c’sle reference set on 2028-02-09. All cross-chip benchmarks below were collected in matched-pair runs in the HIL lab using the same input pipeline, same upstream preprocessing, and the same downstream consumer. See the methodology page for the full protocol.

Task
Depth
Parameters
22.6 M
Benchmarked on
5 chips
Deployments
1.8K

Architecture

Encoder + DPT decoder
Inferred from upstream weights · simplified
ImageViT-S/16Hybrid stemConv stageDPT decoderRefinementDisparity

Headline benchmarks

QSnapdragon 8 Gen 3 NPUINT8
11.2ms p50
89 FPS98.5% acc6.9 W
QQualcomm QCS8550INT8
16.2ms p50
62 FPS98.8% acc13.1 W
NNVIDIA Jetson Orin NanoINT8
18.3ms p50
55 FPS98.4% acc12.1 W

Training data

Pretrained on the upstream maintainer’s released checkpoint. Edge-distillation pass uses 2.4M frames from the Fo’c’sle distillation corpus (consented public data + opt-in publisher contributions). Quantization-aware fine-tune uses 320K calibration samples drawn from the target task’s eval domain.

  • Pretraining corpus: upstream maintainer release
  • Distillation corpus: 2,400,000 frames
  • Calibration set: 320,000 samples (per task)
  • Eval set: standard benchmark + matched-pair HIL runs