Reputation: 43
I work on an OCR project with Matlab and i found out that there is character sample database named MNIST handwritten digit database. I download the file named train-images.idx3-ubyte
but i have no idea how to use it does anyone one know how to use this file ?
Upvotes: 1
Views: 10329
Reputation: 1
You Can read MNISTImages like this-
trlblid = fopen('train-labels.idx1-ubyte');
trimgid = fopen('train-images.idx3-ubyte');
tslblid = fopen('t10k-labels.idx1-ubyte');
tsimgid = fopen('t10k-images.idx3-ubyte');
% read train labels
fread(trlblid, 4);
numtrlbls = toint(fread(trlblid, 4));
trainlabels = fread(trlblid, numtrlbls);
% read train data
fread(trimgid, 4);
numtrimg = toint(fread(trimgid, 4));
trimgh = toint(fread(trimgid, 4));
trimgw = toint(fread(trimgid, 4));
trainimages = permute(reshape(fread(trimgid,trimgh*trimgw*numtrimg),trimgh,trimgw,numtrimg), [2 1 3]);
% read test labels
fread(tslblid, 4);
numtslbls = toint(fread(tslblid, 4));
testlabels = fread(tslblid, numtslbls);
% read test data
fread(tsimgid, 4);
numtsimg = toint(fread(tsimgid, 4));
tsimgh = toint(fread(tsimgid, 4));
tsimgw = toint(fread(tsimgid, 4));
testimages = permute(reshape(fread(tsimgid, tsimgh*tsimgw*numtsimg),tsimgh,tsimgw,numtsimg), [2 1 3]);
Upvotes: 0
Reputation: 318
You can use the mnistHelper functions from Stanford.
As an example of how to use these functions, you can check the images and labels using the following code:
% Change the filenames if you have saved the files under different names
% On some platforms, the files might be saved as
% train-images.idx3-ubyte / train-labels.idx1-ubyte
images = loadMNISTImages('train-images-idx3-ubyte');
labels = loadMNISTLabels('train-labels-idx1-ubyte');
% We are using display_network from the autoencoder code
display_network(images(:,1:100)); % Show the first 100 images
disp(labels(1:10));
Upvotes: 2
Reputation: 19880
Those files you download from MNIST database are binary files. You can find their format on MNIST website: http://yann.lecun.com/exdb/mnist/
Use low-level file I/O functions in MATLAB like fopen, fclose, fread, fseek, etc to read the files following their format.
You can also try to use readMNIST function from FileExchange. I don't have any experience with it, and some users seem to have some problems with it, but you can see the code and debug it.
Upvotes: 2