depth_anything
DA3-BENCH
DA3-BENCH: Depth Anything 3 Evaluation Benchmark
This repository contains processed benchmark datasets for evaluating Depth Anything 3 depth estimation and visual geometry models. The datasets are provided in a convenient, ready-to-use format for research and evaluation purposes.
About Depth Anything 3
Depth Anything 3 (DA3) is a state-of-the-art model that predicts spatially consistent geometry from an arbitrary number of visual inputs, with or without known… See the full description on the dataset page: https://huggingface.co/datasets/depth-anything/DA3-BENCH.DA-2K
DA-2K Evaluation Benchmark
Introduction
DA-2K is proposed in Depth Anything V2 to evaluate the relative depth estimation capability. It encompasses eight representative scenarios of indoor, outdoor, non_real, transparent_reflective, adverse_style, aerial, underwater, and object. It consists of 1K diverse high-quality images and 2K precise pair-wise relative depth annotations.
Please refer to our paper for details in constructing this benchmark.
Usage
Please… See the full description on the dataset page: https://huggingface.co/datasets/depth-anything/DA-2K.middle_depth_anything_kittiDepth-Anything-V2-Datasets
Depth-Anything-V2 Datasets
Depth-Anything-V2 GitHub Repository
Each row is one image triplet (original + greyscale + colourised variants).
Column
Type
Description
id
string
Base image ID e.g. 0000001
id_original
image
Original image
id_gray
image
Greyscale variant (null if absent)
id_color
image
Colourised variant (null if absent)
Example
Original
Colourised DepthGreyscale Depth
Metadata
[
{
"_id": {… See the full description on the dataset page: https://huggingface.co/datasets/ssws3/Depth-Anything-V2-Datasets.ImageNet_RGBD_DepthAnything
ImageNet with Depth generated by DepthAnything ViT-L
DepthAnything_v3_test
Blind Spots of Frontier Models: Depth Anything v3
Dataset Description
Curated by: Pilot Khadka
Task: Monocular Depth Estimation, Failure Case Documentation
Model Evaluated: Depth Anything v3 GIANT (1.13B parameters)
Colab Notebook: https://colab.research.google.com/drive/1rPpE8ua1kmxgd_G1lalY94kEJ0NsV27B#scrollTo=4YqFPtiKNarC
Overview
This dataset documents failure cases of modern frontier vision models, focusing on monocular depth estimation.
The goal of this project… See the full description on the dataset page: https://huggingface.co/datasets/pilot4532/DepthAnything_v3_test.