Comments (2)
The released models are the general-purpose models that are affine invariant. They are arbitrarily scaled with respect to the true metric depth, so you can't directly evaluate RMSE on these models. They also predict inverse depth, not depth, which explains the inverse relationship that you observe.
The RMSE numbers in the the paper refer to models that have been finetuned on the respective datasets. I've just updated the code and uploaded the weights. You can download the weights for KITTI here:
https://github.com/intel-isl/DPT/releases/download/1_0/dpt_hybrid_kitti-cb926ef4.pt
Then run with
python run_monodepth.py -t dpt_hybrid_kitti --kitti_crop
from dpt.
Thank you. This was very helpful. I was able to reduce it to an RMSE of 2.80
Perhaps you might know how to reduce it further?
from dpt.
Related Issues (20)
- RuntimeError: view size is not compatible with input tensor's size HOT 2
- Evaluation for MiDaS v21 HOT 1
- Question about the pos_embed
- Question about finetuning HOT 1
- kitti gt interpolated for training? HOT 1
- Running on large images
- Depth output unit HOT 2
- Add pillow in requirements.txt HOT 1
- Questions about the BlendedMVS datasets used in DPT model
- confidence of the prediction result
- Trying to understand the Dense Prediction Transformer architecture
- Metric Depth from Kitti model HOT 1
- Errors while tracing dpt_beit_large_384.pt
- what is the output metric of the model ? HOT 1
- Thanks and is there a plan to release the dpt-base pretrained model?
- Semantic segmentation inference
- About the datasets for metric depth estimation
- Obataining negative depth values
- DPT_Large Artifacts
- KITTI link is broken HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from dpt.