Ай бұрын
Ай бұрын
2 ай бұрын
2 жыл бұрын
3 жыл бұрын
3 жыл бұрын
3 жыл бұрын
3 жыл бұрын
4 жыл бұрын
4 жыл бұрын
Пікірлер
@shinn-tyanwu4155 18 күн бұрын
Thanks thanks 😊😊
@svenbode7896 25 күн бұрын
Hi Dennis, do I need at first to install pytorch and torchvision in jupyter?
@DennisMadsen 18 күн бұрын
Yes. All the dependencies will need to be installed in your python interpreter. Just use "pip install packagename" for that.
@DennisMadsen Ай бұрын
Also, have a look at the accompanying blog post where all the code snippets can be found: dennismadsen.me/posts/how-to-shape-model-part7
@loanelegall845 Ай бұрын
Hi, I'm Loane Le Gall, I'm studying bioengineering of tissues and implants at the university of aix-marseille, I really like your pedagogy on to discuss this topic so thank you!!! So I'm commenting on this video because for a project I'm trying to develop an ssm of the scapula to allow estimation of the premordbid anatomy of the glenoid so I wanted to know if you were going to address this problem? I've managed to build the ssm, but I'm having trouble with the next steps: how do you estimate the premorbid anatomy? do you have to cut out the damaged part yourself and then reconstruct it using the ssm? I'm a bit lost... I hope you can help me... (I know the scalismo tutorials but they don't help me with this part...) thank you in advance have a nice day :)
@DennisMadsen Ай бұрын
Hi Loane, glad you find the videos useful. If I understand correct, then you have a SSM of the scapula bone and now you want to analyse some scapula bones with a damaged part and try to figure out how they might have looked? If this is the case, remember to only build the SSM on healthy scalula bones - otherwise it it also able to represent damaged bones. Again, correct me if I'm wrong, but now what we want is a classical shape completion. And there you are correct - cut out the damaged part is by far the easiest approach - alternatively you would have some algorithm to detect these "outliers". For the completion I would do this in 3 steps as explained in the video: 1. Cut out a part of one of the meshes used to build your shape model and try to complete the shape using the SSM. Note here that your "correspondence estimation function" should probably be reversed, i.e. finding correspondences from the target to the model. 2. Cut out a part of one of the meshes from a test set. This should be in correspondence with the ones used to build the model. One option is to use all but 1 mesh to build the model, and try to reconstruct the last one. 3. And finally, after you have validated that the code works on the training set, on a test set in correspondence, now you can try out on damaged anatomies where you have to cut out the damaged part. For the manual cutting I would just use meshmixer or something like that. Hope this helps :)
@ryanharvey485 Ай бұрын
'Promosm'
@DennisMadsen Ай бұрын
Also, have a look at the accompanying blog post where all the code snippets can be found: dennismadsen.me/posts/how-to-shape-model-part6
@Beh3D Ай бұрын
model fit generator is now deprecated in tensorflow and when I try to use model.fit the return zip line returns a error regarding the zip object and model.fit, any ideas on how to update the code to fit with the newer versions of python and tensorflow?
@DennisMadsen Ай бұрын
Also, have a look at the accompanying blog post where all the code snippets can be found: dennismadsen.me/posts/how-to-shape-model-part5
@DennisMadsen Ай бұрын
Also, have a look at the accompanying blog post where all the code snippets can be found: dennismadsen.me/posts/how-to-shape-model-part4
@DennisMadsen Ай бұрын
Also, have a look at the accompanying blog post where all the code snippets can be found: dennismadsen.me/posts/how-to-shape-model-part3/
@DennisMadsen 2 ай бұрын
Also, have a look at the accompanying blog post where all the code snippets can be found: dennismadsen.me/posts/how-to-shape-model-part2/
@DennisMadsen 2 ай бұрын
Also, have a look at the accompanying blog post where all the code snippets can be found: dennismadsen.me/posts/how-to-shape-model-part1/
@svenbode7896 2 ай бұрын
Greta Video, best explenation found! Which Jupyter do you use here? Jupyter Notebook, Jupyter Lab or JupyterHub?
@DennisMadsen 18 күн бұрын
I used jupyter notebook. However, more recently started using VS code instead to run the notebooks.
@svenbode7896 2 ай бұрын
Hi Dennis, great tutorial. One question: I installed anaconda3 and pytorch as an admin user and I can run python and import torch, then switched to a non admin user, can run conda, activate an environment and when I import torch in python as the other user, I get an error "ModuleNotFoundError: No module named 'torch'". How can I access the pytorch installation from conda from all users?
@mariyanamadsen9558 2 ай бұрын
🎯🎯🎯
@mybankbalanceis0 3 ай бұрын
what is the windows equivalent of which python command?
@DennisMadsen 2 ай бұрын
I'm not a Windows user myself, but I think `where python` should work.
@shinchannohara3927 3 ай бұрын
will the same code with num-out 200 work for 200 class classification with such great accuracy
@leratomokoena5783 5 ай бұрын
Oh Dennis it is so nice to see how far you have come. Remembering you as the kindest teen you were at kk-electronics aka kk-windsolutions. Best of luck in the future Granny Evelyn....still at KK😉
@BudgiePanic 7 ай бұрын
nice
@user-cc4rq5py1h 8 ай бұрын
unable to train my data
@ibrahelsheikh 8 ай бұрын
very useful
@ibrahelsheikh 8 ай бұрын
very clear
@arpanmaity953 9 ай бұрын
when i am going to train vgg model with 38 classes then that error occur: RuntimeError: Given groups=1, weight of size [64, 3, 3, 3], expected input[2, 38, 224, 224] to have 3 channels, but got 38 channels instead, when i use summarytool, how to slove the error
@rehabhashim8485 10 ай бұрын
Thank you its very good and understandable explaniation
@henoknigatu7121 10 ай бұрын
the image loading and preprocessing you used is a lot easier 👌than the usual way used in pytorch
@KavitaC-bo3fv 11 ай бұрын
Thank you so much for the video. I'm actually trying with UNET for CT volumes. How can I get the image path and mask path of img and mask in show_dataset() and also filename.
@Ruby-jq7ty Жыл бұрын
mine keeps saying select path to conda executable and wont let me proceed does anyone know where I went wrong?
@wurz_alex Жыл бұрын
Hi!! I'm struggling in trying to apply this procedure. I'm using Manjaro (Linux distro) and the problem is that when I get to copy and paste the path to the python interpreter, it is not accepted as it is not seen as a "conda executable". I think it has something to do with the installation of conda or of python itself, how can I solve? Thank you a lot!
@quckychang3486 Жыл бұрын
why I don't have m1? just the base. help!
@user-ql1im6ew7i Жыл бұрын
We need more videos
@user-ql1im6ew7i Жыл бұрын
Why you stop
@tudoronrec Жыл бұрын
Thank you for the information!
@prasunsanki Жыл бұрын
"data/slices/test/mask/img" as per the the data path if I do not use ..../img sub folder after mask or img it should show me figures, but it is showing "Found 0 images belonging to 0 classes." Can you please explain? Thank you in advanced.
@preethireddy5015 Жыл бұрын
U got any solution for this issue?
@wanderXP Жыл бұрын
@@preethireddy5015 No
@user-ot6yk6ie2f Жыл бұрын
i can save this model as usual (by using alexnet) and use it with other model on open cv right?
@kvdiatpune8753 Жыл бұрын
Thanks ,nicely explained
@RishiRaj-hu9it Жыл бұрын
Thanks a lot.. helped me a lot as a beginner
@juandiegoescobarlondono6095 Жыл бұрын
Thank you.
@jeffz7310 Жыл бұрын
you deserve at least ten times more subscriptions
@Baby_Penguin01 Жыл бұрын
thank you bro, was trying to use aesara in pycharm for uni and could only get it working manually through terminal but not when running code through a file. you have saved me a lot of time and manual line by line coding, mega nice
@lucasantagata4551 Жыл бұрын
Did you train your model on CPU or GPU?
@javeriaehsan369 Жыл бұрын
Hello! I have to transfer learning the alexnet for mnist dataset. I have problem that how to find mean and std for that.
@tecknowledger Жыл бұрын
😄 It works!!! Thank you!!!
@raferafe-ui5oh Жыл бұрын
Thanks for the quick succinct explanation. Beginner coder here that was migrating machines and had reconfigured environment three times on the new machine and I could not get pycharm to see any of the environments - 3 min video and done Thanks again
@TimurDE Жыл бұрын
Thank you!!
@MrAlasli Жыл бұрын
Hi Dennis Could you please give us a tutorial on how to do 3D image classification. In particular, classification from the STL files you stored. Great work. Thak you
@anumhassan1422 Жыл бұрын
What if the weights get started negative with increasing iterations?
@redina5257 Жыл бұрын
great video! I want to ask why do we only slice the volume in the X and Y directions? what about the Z direction?
@monteirodelprete6627 Жыл бұрын
Same thing I thought. Consequently It seems that the training is done only on x-slices and y-slices.
@4dimaging619 Жыл бұрын
Hello Dennis, I am working with your GINGR library and use your probablistic ICP algorithm. However, I run into a weird issue. I have two projects based on your GIGR github repo, One is the cleaned version of the other. However, whenever I try to start the probablistic ICP algorithm it runs fine in one of them and the other one get stuck after I run "configProbabilistic.ICP" from your demo configuration. I dont get an error, it just wont show me the iterations in the terminal or UI (so nothing happens) and the script does not finish. The data loads fine and the weird thing is that I exactly copied the code from the other project where it just works (and it is not dependent on scripts that I wrote myself). Therefore, I expect the cause to be somewhere in the project setup. Did you ever experience this before or do you maybe have an idea what causes it?
@SaadShah1133 Жыл бұрын
Great video
@mr_pip_ Жыл бұрын
👍👍👍