-
-
Notifications
You must be signed in to change notification settings - Fork 2.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Example of bag of visual words (BoVW) using SIFT and a RandomForestClassifier #6126
Comments
Hello @glemaitre, Yay, thanks for sharing! We currently have only one gallery example showcasing SIFT; I think your addition would be very valuable. Would you have time to submit your example to the gallery? I think it would fit well under ./doc/examples/applications/. |
I can find time to submit a PR. I am just wondering if there is already a dataset (or a fetcher) in scikit-image that I could reuse? |
I guess you could use the images (which are based on Otherwise, I was looking at http://host.robots.ox.ac.uk/pascal/VOC/ but I can't find any licensing info regarding the datasets... |
There is one existing face detection demo using a small faces vs. non-faces dataset in combination with scikit-learn That ones uses Haar-like features instead of SIFT. could be interesting to compare the performance of the two? I'm not sure that we have any other bundled datasets that could be used. We do store some data externally from this repository at https://gitlab.com/scikit-image/data and it only gets downloaded as needed. |
Yep I can have a look at this dataset.
Now that I see this example, I think that at this time I wanted to reproduce a similar approach to the Haar cascade of Viola and Jones. I could modify a bit this example to be a bit more readable indeed. Basically with BoVW, could be modified and take different detectors/descriptors and then we could make a comparison. |
Hey, there hasn't been any activity on this issue for more than 180 days. For now, we have marked it as "dormant" until there is some new activity. You are welcome to reach out to people by mentioning them here or on our forum if you need more feedback! If you think that this issue is no longer relevant, you may close it by yourself; otherwise, we may do it at some point (either way, it will be done manually). In any case, thank you for your contributions so far! |
I'm super happy to have seen SIFT released recently. It recalls some memory from my computer vision course pre-deep-learning era :)
I quickly drafted (it should be double-checked for bugs) such a scikit-learn compatible transformer that uses SIFT and can be integrated with scikit-learn:
Let me know if it is in the scope of scikit-image and if you would be interested in such an example.
Here, I am using a custom fetcher (see below) and I did not really tune the example to reduce the number of descriptors so it is quite slow.
Also, there is a possibility to make a BoVW more generic that take any detector/descriptor instead of SIFT. I don't know if such a class is indeed of broader interest as well.
Code for fetching data:
The text was updated successfully, but these errors were encountered: