open source pkg v1
This commit is contained in:
Vijay Yadev
202
pkg/OpenFace/matlab_version/demo/face_video_demo.m
Normal file
202
pkg/OpenFace/matlab_version/demo/face_video_demo.m
Normal file
@@ -0,0 +1,202 @@
|
||||
clear
|
||||
addpath('../PDM_helpers/');
|
||||
addpath(genpath('../fitting/'));
|
||||
addpath('../models/');
|
||||
addpath(genpath('../face_detection'));
|
||||
addpath('../CCNF/');
|
||||
|
||||
%%
|
||||
vid_dir = '../../samples/';
|
||||
vids = cat(1, dir([vid_dir, '*.avi']), dir([vid_dir, '*.wmv']));
|
||||
|
||||
%%
|
||||
verbose = true;
|
||||
record = true;
|
||||
|
||||
%% loading the patch experts
|
||||
% Default OpenFace landmark model, using CE-CLM patch experts
|
||||
[patches, pdm, clmParams, early_term_params] = Load_CECLM_general();
|
||||
|
||||
% faster but less accurate
|
||||
%[patches, pdm, clmParams] = Load_CLNF_general();
|
||||
|
||||
% even faster but even less accurate
|
||||
%[patches, pdm, clmParams] = Load_CLM_general();
|
||||
|
||||
% load the face validator and add its dependency
|
||||
load('../face_validation/trained/faceCheckers.mat', 'faceCheckers');
|
||||
addpath(genpath('../face_validation'));
|
||||
od = cd('../face_validation/');
|
||||
setup;
|
||||
cd(od);
|
||||
|
||||
% Setup the face detector (remove the setup mconvnet if not using
|
||||
% MatConvNet)
|
||||
setup_mconvnet;
|
||||
addpath('../face_detection/mtcnn/');
|
||||
|
||||
|
||||
%%
|
||||
for v=1:numel(vids)
|
||||
% load the video
|
||||
vr = VideoReader([vid_dir, vids(v).name]);
|
||||
|
||||
[~,fname,~] = fileparts(vids(v).name);
|
||||
|
||||
if(record)
|
||||
if(~exist('./tracked_vids', 'file'))
|
||||
mkdir('tracked_vids');
|
||||
end
|
||||
writerObj = VideoWriter(sprintf('./tracked_vids/%s.avi', fname));
|
||||
open(writerObj);
|
||||
end
|
||||
|
||||
det = false;
|
||||
initialised = false;
|
||||
|
||||
nFrames = vr.NumberOfFrames;
|
||||
% Read one frame at a time.
|
||||
|
||||
all_local_params = zeros(nFrames, numel(pdm.E));
|
||||
all_global_params = zeros(nFrames,6);
|
||||
|
||||
for i = 1 : nFrames
|
||||
|
||||
% if this version throws a "Dot name reference on non-scalar structure"
|
||||
% error change obj.NumberOfFrames to obj(1).NumberOfFrames (in two
|
||||
% places in read function) or surround it with an empty try catch
|
||||
% statement
|
||||
image_orig = read(vr, i);
|
||||
if((~det && mod(i,4) == 0) || ~initialised)
|
||||
|
||||
% Face detection
|
||||
[bboxs] = detect_faces(image_orig, 'mtcnn');
|
||||
|
||||
% If MTCNN detector not available, can use the cascaded regression one
|
||||
% [bboxs] = detect_faces(image_orig, 'cascade');
|
||||
|
||||
if(~isempty(bboxs))
|
||||
|
||||
% Pick the biggest face for tracking
|
||||
[~,ind] = max(bboxs(:,3) - bboxs(:,1));
|
||||
bbox = bboxs(ind,:);
|
||||
|
||||
% Discard overly small detections
|
||||
if(bbox(3) - bbox(1) > 40)
|
||||
|
||||
% Either infer the local and global shape parameters
|
||||
% from the detected landmarks or just using the
|
||||
% bounding box
|
||||
|
||||
num_points = numel(pdm.M) / 3;
|
||||
|
||||
M = reshape(pdm.M, num_points, 3);
|
||||
width_model = max(M(:,1)) - min(M(:,1));
|
||||
height_model = max(M(:,2)) - min(M(:,2));
|
||||
|
||||
a = (((bbox(3) - bbox(1)) / width_model) + ((bbox(4) - bbox(2))/ height_model)) / 2;
|
||||
|
||||
tx = (bbox(3) + bbox(1))/2;
|
||||
ty = (bbox(4) + bbox(2))/2;
|
||||
|
||||
% correct it so that the bounding box is just around the minimum
|
||||
% and maximum point in the initialised face
|
||||
tx = tx - a*(min(M(:,1)) + max(M(:,1)))/2;
|
||||
ty = ty + a*(min(M(:,2)) + max(M(:,2)))/2;
|
||||
|
||||
% visualisation
|
||||
g_param_n = [a, 0, 0, 0, tx, ty]';
|
||||
|
||||
l_param_n = zeros(size(pdm.E));
|
||||
|
||||
% If tracking has not started trust the detection
|
||||
if(~initialised)
|
||||
g_param = g_param_n;
|
||||
l_param = l_param_n;
|
||||
det = true;
|
||||
initialised = true;
|
||||
else
|
||||
% If tracking has already started double check the
|
||||
% detection
|
||||
shape_new = GetShapeOrtho(pdm.M, pdm.V, params, g_param_n);
|
||||
|
||||
dec = face_check_cnn(image, shape_new, g_param, faceCheckers);
|
||||
|
||||
if(dec < 0.5)
|
||||
det = true;
|
||||
g_param = g_param_n;
|
||||
l_param = l_param_n;
|
||||
else
|
||||
det = false;
|
||||
end
|
||||
end
|
||||
|
||||
end
|
||||
end
|
||||
end
|
||||
if(size(image_orig,3) == 3)
|
||||
image = rgb2gray(image_orig);
|
||||
else
|
||||
image = image_orig;
|
||||
end
|
||||
|
||||
d_image = [];
|
||||
|
||||
if(initialised)
|
||||
[shape,g_param,l_param,lhood,lmark_lhood,view_used] = Fitting_from_bb(image, d_image, bbox, pdm, patches, clmParams, 'gparam', g_param, 'lparam', l_param);
|
||||
all_local_params(i,:) = l_param;
|
||||
all_global_params(i,:) = g_param;
|
||||
|
||||
dec = face_check_cnn(image, shape, g_param, faceCheckers);
|
||||
|
||||
if(dec < 0.5)
|
||||
clmParams.window_size = [19,19; 17,17;];
|
||||
clmParams.numPatchIters = 2;
|
||||
det = true;
|
||||
else
|
||||
clmParams.window_size = [21,21; 19,19; 17,17;];
|
||||
clmParams.numPatchIters = 3;
|
||||
det = false;
|
||||
end
|
||||
end
|
||||
|
||||
if(verbose)
|
||||
|
||||
try
|
||||
if(max(image_orig(:)) > 1)
|
||||
imshow(double(image_orig)/255, 'Border', 'tight');
|
||||
else
|
||||
imshow(double(image_orig), 'Border', 'tight');
|
||||
end
|
||||
axis equal;
|
||||
hold on;
|
||||
|
||||
if(initialised)
|
||||
plot(shape(:,1), shape(:,2),'.r','MarkerSize',20);
|
||||
plot(shape(:,1), shape(:,2),'.b','MarkerSize',10);
|
||||
end
|
||||
hold off;
|
||||
drawnow expose;
|
||||
pause(0.01);
|
||||
|
||||
if(record)
|
||||
frame = getframe;
|
||||
writeVideo(writerObj,frame);
|
||||
end
|
||||
|
||||
catch warn
|
||||
fprintf('%s', warn.message);
|
||||
end
|
||||
end
|
||||
|
||||
end
|
||||
if(record)
|
||||
close(writerObj);
|
||||
end
|
||||
|
||||
close all;
|
||||
|
||||
experiments.local_params = all_local_params;
|
||||
experiments.global_params = all_global_params;
|
||||
|
||||
end
|
||||
Reference in New Issue
Block a user