Merge pull request #58 from MATLAB-Community-Toolboxes-at-INCF/enpack…

…-datastore

Relocate custom datastore to root namespace

datastore/DeepInterpolationDataStore.m → +deepinterp/Datastore.m

@@ -1,12 +1,12 @@
-classdef DeepInterpolationDataStore < matlab.io.Datastore & ...
+classdef Datastore < matlab.io.Datastore & ...
                                       matlab.io.datastore.Subsettable & ...
                                       matlab.io.datastore.Shuffleable
-    %DEEPINTERPOLATIONDATASTORE - custom datastore for DeepInterpolation-MATLAB
+    % DATASTORE - custom datastore for DeepInterpolation-MATLAB
     %
     % This custom datastore provides flanking frames and the center frame 
     % upon one read.
     %
-    % dids = DeepInterpolationDataStore(pathToTiffFile, options);
+    % dids = deepinterp.Datastore(pathToTiffFile, options);
     %
     % Input must be full path to a grayscale, single channel tiff stack of 
     % >'flankingFrames + 2' frames in order to work (this is according to
@@ -36,7 +36,7 @@
     end
 
     methods % begin methods section
-        function myds = DeepInterpolationDataStore(filePath, options)
+        function myds = Datastore(filePath, options)
             arguments
                 filePath
                 options = struct();
@@ -257,4 +257,4 @@ function reset(myds)
 
     end
 
-end
+end

+deepinterp/setup.m

@@ -12,5 +12,5 @@ function setup()
 addpath(fullfile(tbd, "sampleData"))
 addpath(fullfile(tbd, "pretrainedModels"))
 addpath(fullfile(tbd, "examples"))
-addpath(fullfile(tbd, "datastore"))
+
 

README.md

@@ -56,7 +56,7 @@ The principle behind DeepInterpolation is illustrated in the following figure fr
 ![Figure 1a from Lecoq et al. 2021](sampleData/Lecoq_et_al_2021_Fig1a.png). 
 --> 
 
-To predict the frame-of-interest ("predicted frame"), a deep learning network uses data from several preceeding and succeeding video frames. During training, the network is modified so that it produces better and better representations of the predicted frames over several datasets. During inference, the network produces predicted frames that are used in place of the original data. If the signal in the data is well predicted by the information in the preceeding and succeeding frames, then the inferred data contains a reconstruction of the original data where the usual noise that occurs independently on each frame is greatly reduced, because it is not predicted on average. The signal can be inferred with high quality in part because the network can average out the independent noise in the preceeding and succeeding frames to uncover the underlying signal just prior to and just after the signal in the predicted frame. 
+To predict the frame-of-interest ("predicted frame"), a deep learning network uses data from several preceeding and succeeding video frames. During training, the network is modified so that it produces better and better representations of the predicted frames over several datasets. During inference, the network produces predicted frames that are used in place of the original data. DeepInterpolation works well in situations where the signal in the data is well predicted by the information in the preceeding and succeeding frames. In these cases, the inferred data contains a good reconstruction of the underlying signal while the noise that occurs independently on each frame is greatly reduced, because the noise is not predicted on average.
 
 ### Installation
 🚧

tests/.gitignore

@@ -0,0 +1 @@
+

datastore/test/TestDeepInterpolationDataStore.m → tests/TestDeepInterpolationDataStore.m

@@ -18,7 +18,7 @@
 
 %% Test 01: Check if  matlab.io.Datastore is superclass.
 % A deepinterpolation datastore inherits from abstract Datastore
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 assert(isa(dids,'matlab.io.Datastore'));
 
 %% Test 02: Call the read
@@ -30,7 +30,7 @@
 % The first frame that allows that is the 32nd!
 % The combination of training data and center frame is called a "set" here, and the
 % number of sets, following the logic above, is N-Frames - 62.
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 t = read(dids);
 assert(iscell(t));
 assert(numel(t) == 2);
@@ -39,7 +39,7 @@
 
 %% Test 03: Call the read again
 % Standard sequential datastore behavior: advance to "second" center-frame
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 read(dids);
 [t,info] = read(dids);
 assert(iscell(t));
@@ -50,7 +50,7 @@
 
 %% Test 04: Successfully read until no more data
 % The testdatastack should return 38 sets of training frames and reference frame
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 count = 0;
 while(hasdata(dids))
     read(dids);
@@ -59,7 +59,7 @@
 assert(count == 38);
 
 %% Test 05: Call read when out of data
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 try
     read(dids);
 catch ME
@@ -69,12 +69,12 @@
 %% Test 11: Call the readall method
 % Standard sequential datastore behavior: return all data in a nx2 Cell array, for
 % the test dataset n==38
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 allt = readall(dids);
 assert(all(size(allt)==[38 2]));
 
 %% Test 12: Call the readall method when out of data
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 while(hasdata(dids))
   read(dids);
 end
@@ -83,7 +83,7 @@
 
 %% Test 21: Call the reset method before any read
 % Reset to "first" frame (in fact, the 32nd frame in the stack...)
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 reset(dids);
 [t,info] = read(dids);
 assert(iscell(t));
@@ -93,7 +93,7 @@
 assert(info == "set=1/centerframe=32");
 
 %% Test 22: Call the reset method after some reads
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 read(dids);
 read(dids);
 read(dids);
@@ -106,7 +106,7 @@
 assert(info == "set=1/centerframe=32");
 
 %% Test 23: Call the reset method after readall
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 read(dids);
 tall = readall(dids);
 reset(dids);
@@ -118,7 +118,7 @@
 assert(info == "set=1/centerframe=32");
 
 %% Test 24: Call the reset method when out of data
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 while(hasdata(dids))
   read(dids);
 end
@@ -131,32 +131,32 @@
 assert(info == "set=1/centerframe=32");
 
 %% Test 31: Call the progress method before any reads
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 p = progress(dids);
 assert(p==0);
 
 %% Test 32: Call the progress method after readall but before read.
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 readall(dids);
 p = progress(dids);
 assert(p==0);
 
 %% Test 33: Call the progress method after read
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 read(dids);
 p = progress(dids);
 assert((p-0.0270)<1e-4);
 
 %% Test 34: Call the progress method when out of data
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 while(hasdata(dids))
   read(dids);
 end
 p = progress(dids);
 assert((p-1.0)<1e-4);
 
 %% Test 35: Progress produces correct output over all sets
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 idx = 1;
 while(hasdata(dids))
   p(idx) = progress(dids);
@@ -168,7 +168,7 @@
 assert(all(diff(p) > 0), "progress must be monotonically increasing");
 
 %% Test 41: Preview methods returns first set
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 t = preview(dids);
 assert(iscell(t));
 assert(numel(t) == 2);
@@ -177,7 +177,7 @@
 assert((t{1}(323,23,44)-138)<1e-4);
 
 %% Test 42: Preview methods returns first set after reads
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 read(dids);
 read(dids);
 t = preview(dids);
@@ -188,7 +188,7 @@
 assert((t{1}(323,23,44)-138)<1e-4);
 
 %% Test 43: Preview methods returns first set after readall
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 readall(dids);
 t = preview(dids);
 assert(iscell(t));
@@ -198,7 +198,7 @@
 assert((t{1}(323,23,44)-138)<1e-4);
 
 %% Test 44: Preview methods returns first set after reads and reset
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 read(dids);
 read(dids);
 reset(dids);
@@ -210,7 +210,7 @@
 assert((t{1}(323,23,44)-138)<1e-4);
 
 %% Test 45: Preview methods returns first set when out of data
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 while hasdata(dids)
     read(dids);
 end
@@ -222,7 +222,7 @@
 assert((t{1}(323,23,44)-138)<1e-4);
 
 %% Test 46: Read after preview is from correct location
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 read(dids);
 read(dids);
 preview(dids);
@@ -234,13 +234,13 @@
 assert(info == "set=3/centerframe=34");
 
 %% Test 47: Readall after preview works correctly
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 preview(dids);
 allt = readall(dids);
 assert(all(size(allt)==[38 2]));
 
 %% Test 48: Hasdata works correctly after preview
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 read(dids);
 preview(dids);
 assert(hasdata(dids));
@@ -249,22 +249,22 @@
 % A partition of the deepinterpolation datastore is merely pointing to a different
 % startframe and has a smaller number of available sets. All the other behaviors are
 % derived from that
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = partition(dids,10,7);
 assert(isa(subds,'matlab.io.Datastore'));
 
 %% Test 51: Call Partition, check if the partition has correct properties
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = partition(dids,10,7);
 assert(isequal(properties(dids),properties(subds)));
 
 %% Test 52: Call Partition, check if the partition has correct methods
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = partition(dids,10,7);
 assert(isequal(methods(dids),methods(subds)));
 
 %% Test 53: Call Partition, successfully read from a partition
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = partition(dids,10,7);
 [t,info] = read(subds);
 assert(iscell(t));
@@ -274,14 +274,14 @@
 assert(info == "set=1/centerframe=56");
 
 %% Test 54: Call last Partition, read all data
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = partition(dids,13,13);
 while hasdata(subds)
     read(subds);
 end
 
 %% Test 55: Call Partition, check that total number of sets/frames is correct
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 partsets = 0;
 for iii = 1:10
     subds = partition(dids,10,iii);
@@ -290,15 +290,15 @@
 assert(partsets-38 < 1e-4);
 
 %% Test 56: Call partition with partition=1 and index=1 and validate
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = partition(dids,1,1);
 assert(isequal(properties(dids),properties(subds)));
 assert(isequal(methods(dids),methods(subds)));
 assert(isequaln(read(subds),read(dids)));
 assert(isequaln(preview(subds),preview(dids)));
 
 %% Test 57: Partition a partition again and validate
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = partition(dids,2,2);
 subsubds = partition(subds,2,2);
 [t,info] = read(subsubds);
@@ -309,7 +309,7 @@
 assert(info == "set=1/centerframe=61");
 
 %% Test 58: Progress is working correctly in partitions
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = partition(dids,2,2);
 idx = 1;
 clear p
@@ -327,7 +327,7 @@
 %% Test 60: Create a Subset of the datastore and validate
 % deepinterpolation datastore is subsettable, again this just assigns the correct
 % startframe and number of available sets, but points to the same file
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = subset(dids,5:20);
 [t,info]=read(subds);
 assert(iscell(t));
@@ -337,7 +337,7 @@
 assert(info == "set=1/centerframe=36");
 
 %% Test 61: Progress is working correctly in subsets
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 subds = subset(dids,5:10);
 idx = 1;
 clear p
@@ -354,27 +354,27 @@
 
 %% Test 71: Fail on wrong image size
 try
-    dids = DeepInterpolationDataStore(WRONGSIZETESTSTACKFULLFILE); %#ok<NASGU>
+    dids = deepinterp.Datastore(WRONGSIZETESTSTACKFULLFILE); %#ok<NASGU>
 catch EM
     assert(strcmp(EM.message,'Actual frame size is not equal to specified outputFrameSize'),"Fail on wrong tiff size");
 end
 
 %% Test 72: Read data with automatic resize
 options.doAutoResize = true;
-dids = DeepInterpolationDataStore(WRONGSIZETESTSTACKFULLFILE, options);
+dids = deepinterp.Datastore(WRONGSIZETESTSTACKFULLFILE, options);
 t = read(dids);
 assert(iscell(t));
 assert(numel(t) == 2);
 assert(all((size(t{1})==[512 512 60])));
 assert(all((size(t{2})==[512 512])));
 
 %% Test 81: Shuffling returns randomized datastore without error
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 shds = shuffle(dids);
 assert(isa(shds,'matlab.io.Datastore'));
 
 %% Test 82: can read sequentially from shuffled datastore
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 shds = shuffle(dids);
 count = 0;
 while(hasdata(shds))
@@ -384,7 +384,7 @@
 assert(count == 38);
 
 %% Test 83: can read all from shuffled datastore
-dids = DeepInterpolationDataStore(TESTSTACKFULLFILE);
+dids = deepinterp.Datastore(TESTSTACKFULLFILE);
 shds = shuffle(dids);
 allt = readall(shds);
-assert(all(size(allt)==[38 2]));
+assert(all(size(allt)==[38 2]));