Comments (13)
Thanks for the report. Do you feel like submitting a pull request? Since this is a simple change, you could to navigate to https://github.com/JuliaImages/ImageFeatures.jl/blob/master/docs/src/tutorials/object_detection.md and click on the pencil in the upper right to edit the file.
from imagefeatures.jl.
The tutorial uses this pedestrian dataset - http://cbcl.mit.edu/software-datasets/PedestrianData.html. The download link isn't working right now and I can't find dataset anywhere else.
Thanks for reporting this. I will update the tutorial to use another dataset (by this weekend).
from imagefeatures.jl.
I have found the data here:
Maybe you could simply add it in the repo and load it from the repo. When compressed its only some MB.
from imagefeatures.jl.
An even better idea would be to add the data in MLDatasets.jl so other people can use it independently of ImageFeatures easily.
from imagefeatures.jl.
Also, there might be a small typo in the get_data.bash
file. The name of the Julia file in the bash file(line number 11) is preprocess_data.jl
whereas the actual file is preprocess.jl
. I could have sent a PR, but this is in the downloaded Tutorial.zip folder.
from imagefeatures.jl.
@timholy Should I add the contents of Tutorial.zip to this repository itself? Since its on google drive currently, only I can update the content of the tutorial.
from imagefeatures.jl.
I'm a little reluctant to add a 10MB file to the repository (and once added it becomes a bottleneck on the git history even if we delete it later). What about adding a download script to MLDatasets.jl as suggested by @davidbp? It would be a bit like TestImages.jl.
from imagefeatures.jl.
A simple ad-hoc solution would be to use https://github.com/oxinabox/DataDeps.jl from @oxinabox directly. That package will be the future backend of MLDatasets (see JuliaML/MLDatasets.jl#12), so moving the dataset code to MLDatasets would be possible in the future. The bonus of using DataDeps directly is that it requires a lot less effort to do (tests and documentation) than I am advocating for MLDatasets.
from imagefeatures.jl.
I think this is a good use case for it yes.
I've used DataDeps.jl in a couple of quick scripts that had some data depenancy and it worked well and cleanly.
It is still v0.x.y so feedback on a design level would still be appreciated.
I want it to be good for this use case -- when you want the data as a kind of one-off, rather than wanting to go the full level of adding it to MLDatasets.jl etc. (as well as for the other usecase as a backend downloader for MlDatasets.jl etc)
If it is not working for this or if the code ends up ugly, then let me know.
(I am currently on holidays so am not responding too quick on Github right now, but I'll be back off next week)
from imagefeatures.jl.
I'm getting a data error while trying to read the non-human images (source provided by @davidbp ) from the tutorial (see #48 (comment)). In trying to track down an alternate source of the data I came across a possibility at http://poggio-lab.mit.edu/PedestrianData.html however this yields a 404. I wrote to the contact mentioned on the lab page (dlees) but this address is no longer functioning.
from imagefeatures.jl.
You can download the original dataset by using the wayback machine. The following link appears to be downloading for me.
from imagefeatures.jl.
@zygmuntszpak As far as I can see that link points to the human images, not the non_human. So the solution was to take the not_human images I have and resize them. The tutorial now runs to the end.
from imagefeatures.jl.
@zygmuntszpak The link works but it's not the same data. This link contains 924 images not 2500 images. Besides there are no Negative samples. I don't know where to download the exact same data from the tutorial.
from imagefeatures.jl.
Related Issues (20)
- Build homography from array of matched points HOT 3
- HOG documentation
- GMS: Grid-based Motion Statistics for Fast, Ultra-robust Feature Correspondence
- Rationale for Vector of CartesianIndex{2} for output 'matches' of match_keypoints HOT 16
- Dependency on Images.jl directly HOT 5
- Error while trying to run the HOG example HOT 3
- HOG not working HOT 20
- Improving Hog allocations / Making a fast HOG version HOT 2
- Is it on the roadmap to support 1.0 in the near feature HOT 7
- Are there any plans to add sift and surf ?? HOT 12
- docstring missing HOT 3
- corner.jl: ERROR: MethodError: no method matching dotc(::RGBA{Float64}, ::RGBA{Float64}) HOT 1
- hough_circle_gradient: centers could hold 0 in CartesianIndex
- brief.jl:There is an error in the range HOT 3
- Getting the votes from the Hough transform HOT 5
- Images is downgraded
- TagBot trigger issue HOT 14
- coords_spatial not defined HOT 3
- Symmetry transform
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 imagefeatures.jl.