Tejus Gupta

@timholy I implemented the algorithm described in [A coarse-to-fine algorithm for fast median filtering of image data with a huge number of levels by Alparone et al.](https://www.sciencedirect.com/science/article/pii/016516849490121X) and got some...

@hari-sikchi You should also read [this paper](https://nomis80.org/ctmf.pdf). I think it would be best to write separate functions for cases where the color values are discrete and when they are continuous....

@timholy Are there functions to compute union and difference for CartesianRange? I think it would be easier to keep track of pixels to add/subtract if we could simply compute union/difference...

I used the pretrained model for segmentation i.e. `python demo.py --input_image data/demo/test_image.png` [Test Image Link](https://user-images.githubusercontent.com/17504447/31215340-bae2dfdc-a9cc-11e7-82a5-846ab87ba092.png)

I did some benchmarking for NearestNeighbors.jl. For the same algorithm (KDTree), FLANN has about 10x better performance. I think this is because FLANN has a system for automatically choosing the...

I meant to say that I tested KristofferC/NearestNeighbors.jl and it was 10x slower that FLANN.jl. [https://github.com/johnmyleswhite/NearestNeighbors.jl](https://github.com/johnmyleswhite/NearestNeighbors.jl) only supports brute force search right now. The README says that they plan to...

Understood. I am trying to fix the build scripts for FLANN.jl.

Support for 64-bit windows has been added to FLANN.jl - https://github.com/wildart/FLANN.jl/issues/3. Would it be ok to use [NearestNeighbors.jl](https://github.com/KristofferC/NearestNeighbors.jl) for windows 32 bit and use the more optimized FLANN.jl for other...

@timholy This is ready for review. Comparison of current and previous matcher - ``` using ImageFeatures, Images, NearestNeighbors img1 = rand(Gray{N0f8}, 100, 100); img2 = rand(Gray{N0f8}, 100, 100); keypoints_1 =...

Since flann gives approximate results, the test case on line 165 of brief.jl fails. 2 out of the 267 matches returned are incorrect. ``` @test sum(m[1] + CartesianIndex(100, 200) ==...