add support for 1-D and 3-D images, Logger interface, TBImage Wrapper, Tests, Examples

[shashikdm]

Added L, CL, LC formats. added respective tests also

[codecov-io]

Codecov Report

Merging #24 into master will decrease coverage by 3.35%.
The diff coverage is 92.85%.

@@            Coverage Diff             @@
##           master      #24      +/-   ##
==========================================
- Coverage   82.66%   79.31%   -3.36%     
==========================================
  Files          24       24              
  Lines         323      290      -33     
==========================================
- Hits          267      230      -37     
- Misses         56       60       +4

Impacted Files	Coverage Δ
src/TensorBoardLogger.jl	100% <ø> (ø)	⬆️
src/logger_dispatch.jl	69.23% <100%> (+16.59%)	⬆️
src/Loggers/LogImage.jl	100% <100%> (ø)	⬆️
src/logger_dispatch_overrides.jl	69.23% <79.16%> (+15.89%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update fe1276a...103d1dc. Read the comment docs.

[shashikdm]

Highlights:

• seperate image_summary for {AbstractArray, format} and {AbstractArray{:<Colorant}}(image objects)
• autmatically detect format from image objects, must provide format for number arrays
• automatic dispatch of image_summary for image objects when using logger interface
• TBImage to be used only for numerical arrays and takes format as an argument
• TBImages also takes format as argument and creates TBImage objects for each

[shashikdm]

TODO:

• Refactoring
• more test coverage

[oxinabox]

I was trying to workout how to cut down some of the common code in image_summary
More iformation can be encoded into the enum

# Note: we use the 100s place of the enum to represent the numer of dimensions
# And the 10s place to represent which dimension is the observation (N), 0 for no obseration dimension 
@enum ImageFormat begin
    L=100; 
    CL=200; LC; HW; WH; NL=210; LN=220;  
    HWC=300; WHC; CHW; CWH; NCL=310; NHW; NWH; NLC; CLN=330; LCN; HWN; WHN;  
    NHWC=410; NWHC; NCHW; NCWH; HWCN=440; WHCN; CHWN; CWHN; 
end

expected_ndims(x::ImageFormat) = Int(x) ÷ 100  # get the 100's digit
obs_dim(x::ImageFormat) = (Int(x) % 100) ÷ 10   # get the 10's digit

[shashikdm]

I was trying to workout how to cut down some of the common code in image_summary
More iformation can be encoded into the enum

# Note: we use the 100s place of the enum to represent the numer of dimensions
# And the 10s place to represent which dimension is the observation (N), 0 for no obseration dimension 
@enum ImageFormat begin
    L=100; 
    CL=200; LC; HW; WH; NL=210; LN=220;  
    HWC=300; WHC; CHW; CWH; NCL=310; NHW; NWH; NLC; CLN=330; LCN; HWN; WHN;  
    NHWC=410; NWHC; NCHW; NCWH; HWCN=440; WHCN; CHWN; CWHN; 
end

expected_ndims(x::ImageFormat) = Int(x) ÷ 100  # get the 100's digit
obs_dim(x::ImageFormat) = (Int(x) % 100) ÷ 10   # get the 10's digit

this is awesome! thank you very much

[shashikdm]

shorter code (+65 −378 ), only about 1/3rd of the original. Greate idea @oxinabox.

[zenna]

Just wanted to add that this is dope and highly useful for the upcoming NIPS deadline.

[PhilipVinc]

Very nice!
I can merge it anytime you want.

I'll also tag another minor release with that.

[shashikdm]

@PhilipVinc you can go ahead and merge this. I suggest waiting a few more days for log_audio and then tag a release. It's almost ready in my branch

[PhilipVinc]

Thanks a lot for your efforts.

I'll tag a release anyway as we can just tag a new one after log_audio is implemented.