jspenmar / monodepth_benchmark Goto Github PK
View Code? Open in Web Editor NEWCode for "Deconstructing Monocular Depth Reconstruction: The Design Decisions that Matter" (https://arxiv.org/abs/2208.01489)
License: Other
Code for "Deconstructing Monocular Depth Reconstruction: The Design Decisions that Matter" (https://arxiv.org/abs/2208.01489)
License: Other
Hi, thanks for the wonderful codebase.
I would like to ask which configuration has the highest accuracy on KITTI Eigen and SYNS-Patches respectively? The corresponding precision of these experimental configurations is not found in readme.
Look forward for your reply, thanks.
Hello, thank you for the amazing work!
I am downloading the dataset following the instructions on https://github.com/jspenmar/monodepth_benchmark/blob/main/api/data/README.md
.
In the Download section of Kitti Depth Benchmark it says to execute
python ../../api/data/export_kitti_depth_benchmark.py # Copy benchmark data to 'kitti_raw_sync'
When I run that I get a ModuleNotFoundError: No module named 'src'
.
If I move to the root and try to run the script as a module I get a KeyError.
How can I solve that?
I am trying to reproduce monodepth2 results using the benchmark. I run
python api/train/train.py -cfg-file cfg/abl_bb/resnet18.yaml --cfg-default cfg/abl_bb/default.yaml --ckpt-dir logs/benchmark/ --name md2 --version 42 --seed 42
where resnet18.yaml
is
net:
depth:
enc_name: 'resnet18'
dataset:
type: 'kitti'
supp_idxs: [ -1, 1]
loader:
batch_size: 36
But, when I evaluate it, my best model is not as good at monodepth2
Ratio: 26.717182107430105
AbsRel: 0.1205853946784904
SqRel: 0.9113600682814631
RMSE: 4.980470443558659
LogRMSE: 0.19857460770280325
$\delta < 1.25$: 0.8655597941752403
$\delta < 1.25^2$: 0.9562950148720958
$\delta < 1.25^3$: 0.9799390811001284
Could you please clarify how model results can be evaluated on syns-patches dataset, since there doesn't seem to be a submission option available on the challenge page on CodaLab?
In api/data/export_kitti_depth_benchmark.py
some syntax do not support in python3.6
and numpy should update 1.21.x ,but python3.6 not support this
mybe there are other error by lib version or pytho version,so if ok , add information of python version and requirements.txt.
Hi @jspenmar, thank you for open-sourcing this project, it is exciting!
Regarding the re-implemented models, will you release the evaluation scripts as well as the pretrained weights for reproduction? Thanks!
Hello,
I am working on a project that involves monocular depth estimation using SYNS-Patches dataset. I would like to request the camera parameters of this dataset, especially intrinsics such as focal length and principal point.
Could you please provide me with these information or point me to where I can find them? I appreciate your help.
Thank you.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.