monodepth2
glasses
Our great sponsors
monodepth2 | glasses | |
---|---|---|
6 | 2 | |
3,974 | 413 | |
1.5% | - | |
0.0 | 1.8 | |
7 months ago | over 1 year ago | |
Jupyter Notebook | Jupyter Notebook | |
GNU General Public License v3.0 or later | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
monodepth2
- Calculation of an absolute depth map from multiple images or videos.
-
Easy to train a monocular (self) supervised depth estimation model?
I've used monodepth2 before and it's great: https://github.com/nianticlabs/monodepth2
-
Sources: Pixel 6 Pro was supposed to launch with face unlock
How can a single camera do that? My experience with computer vision is fairly limited so I'm curious how that would work. My understanding is you need to be able to generate a point map, stereo vision, or some non-CV related method , e.g. radar like the pixel 4. 2D depth estimation can be done with a single camera in somewhat useful way but it's not a secure way (https://github.com/nianticlabs/monodepth2 -- now somewhat similar functionality in OpenCV). Can you expand on what AI the single camera is being combined with that provides security guarantees?
- Can anyone explain the following github code to me. Also itβs my first time using GitHub so Iβm completely lost.
-
Estimating camera height, orientation and field of view from a single monocular image.
I suspect you may have the best success by using monocular depth approaches (for example something like this: https://github.com/nianticlabs/monodepth2).
-
Looking for a fast monocular depth estimation library to use in a Rust project.
After that I have to do the same for Python I think, and then I have to find out how to figure out how to use a library like https://github.com/ialhashim/DenseDepth or https://github.com/nianticlabs/monodepth2 for that GStreamer plugin (or element, still trying to grasp the terminology here)
glasses
-
Are Open-sourced Implementations Sometimes Over-engineered?
Yes, they are. Take with a grain of salt, but researchers (usually) do not know how to code and (or) they don't care to properly share their work. Things that are learned in the first Computer Science bachelor year, like OOP, DRY, packages, good variables/function naming, are apparently not used in ml research. This is why I created my own library (https://github.com/FrancescoSaverioZuppichini/glasses), for me, good code means less time I have to spend working and more free time.
-
[N] Facebook announced a new AI open-source called DeiT (A new technique to train computer vision models)
I have implemented most of the sota models in my library (https://github.com/FrancescoSaverioZuppichini/glasses). These are my 2 cents:
What are some alternatives?
DenseDepth - High Quality Monocular Depth Estimation via Transfer Learning
yolov5 - YOLOv5 π in PyTorch > ONNX > CoreML > TFLite
packnet-sfm - TRI-ML Monocular Depth Estimation Repository
One-Piece-Image-Classifier - A quick image classifier trained with manually selected One Piece images.
cs231n - Note and Assignments for CS231n: Convolutional Neural Networks for Visual Recognition
conformal_classification - Wrapper for a PyTorch classifier which allows it to output prediction sets. The sets are theoretically guaranteed to contain the true class with high probability (via conformal prediction).
torchdyn - A PyTorch library entirely dedicated to neural differential equations, implicit models and related numerical methods
gan-vae-pretrained-pytorch - Pretrained GANs + VAEs + classifiers for MNIST/CIFAR in pytorch.
ZoeDepth - Metric depth estimation from a single image
neuralforecast - Scalable and user friendly neural :brain: forecasting algorithms.
depth-estimate-gui - Depth Estimate GUI - Windows, Mac, Linux
HugsVision - HugsVision is a easy to use huggingface wrapper for state-of-the-art computer vision