def __init__(self, exp_folder, lims_ID):
for file in os.listdir(exp_folder):
# looks for the excel file and makes the directory to it
if file.endswith(".mp4") and file.startswith(lims_ID):
self.directory = exp_folder
self.file_string = os.path.join(exp_folder, file)
self.sb = sb(exp_folder)
self.ep = ep(exp_folder, lims_ID)
self.sv = sv (exp_folder, lims_ID)
self.video_pointer = cv2.VideoCapture(self.file_string)
if os.path.isfile(self.file_string):
self.data_present = True
else:
self.data_present = False
def __init__(self, exp_folder, lims_ID):
for file in os.listdir(exp_folder):
# looks for the excel file and makes the directory to it
if file.endswith(".mp4") and file.startswith(lims_ID):
self.directory = exp_folder
self.file_string = os.path.join(exp_folder, file)
self.sb = sb(exp_folder)
self.ep = ep(exp_folder, lims_ID)
self.sv = sv(exp_folder, lims_ID)
self.video_pointer = cv2.VideoCapture(self.file_string)
if os.path.isfile(self.file_string):
self.data_present = True
else:
self.data_present = False
def show_frame(frame):
cv2.imshow('image', frame)
cv2.waitKey(0)
cv2.destroyAllWindows()
if __name__ == '__main__':
lims_ID = ['501560436', '502741583','509904120', '500860585']
final_data = []
y_train = []
feature_data=[]
y_fid=[]
wheel=[]
t= 0
for itm in lims_ID:
exl = ep("C:\Users\mahdir\Desktop\Mahdi files", itm)
data = get_data(exl, itm)
if t == 0:
final_data = data['final_data']
y_train= data['y_train']
feature_data = data['fidget_data']
y_fid = data['y_fid']
wheel = data['wheel']
else:
vector = data['final_data']
final_data = np.vstack((final_data, vector))
y_train = np.concatenate((y_train, data['y_train']))
feature_data = np.concatenate((feature_data, data['fidget_data']))
y_fid = np.concatenate((y_fid, data['y_fid'] ))
wheel = np.concatenate((wheel, data['wheel']))
t += 1
cv2.destroyAllWindows()
if __name__ == '__main__':
# set limsID of video data to train on
# '503412730', '497060401','502741583', '501004031', '500860585', '501560436'
lims_ID = ['502741583', '501004031', '500860585', '501560436']
# initialize training data and data label arrays, as well a a loop counter
y_train = []
y_track = []
feature_data = []
t = 0
# for each limd_ID, get trainign and label data, and combine it with previous lims ID training and label data
for itm in lims_ID:
ex = ep("C:\Users\mahdir\Documents\Allen Projects\Behavior Annotation",
itm)
data = get_data(itm, ex)
if t == 0:
y_train = data['y_train']
y_track = data['y_track']
feature_data = data['feature_data'][0:len(y_train)]
print(itm + ' video done')
print(data['number'])
else:
y_vector = data['y_train']
y_train = np.concatenate((y_train, y_vector))
y_track = np.concatenate((y_track, data['y_track']))
vector = data['feature_data'][0:len(y_vector)]
feature_data = np.vstack((feature_data, vector))
print(itm + ' video done')
def run_whole_video(exp_folder, lims_ID):
#initializes video pointer for video of interest based on lims ID
file_string = get_file_string(exp_folder + '\Videos\\', lims_ID)
video_pointer = cv2.VideoCapture(file_string)
# load feature scalers
scale_wheel = joblib.load(exp_folder + '\Scalers\\' + 'wheel_scale_' +
str(lims_ID) + '.pkl')
scale_frame = joblib.load(exp_folder + '\Scalers\\' + 'frame_scale_' +
str(lims_ID) + '.pkl')
scale_optical = joblib.load(exp_folder + '\Scalers\\' + 'optical_scale_' +
str(lims_ID) + '.pkl')
scale_angle = joblib.load(exp_folder + '\Scalers\\' + 'angle_scale_' +
str(lims_ID) + '.pkl')
# import wheel data
dir = os.path.join(exp_folder + '\Wheel\\', 'dxds' + str(lims_ID) + '.pkl')
wheel = joblib.load(dir)
first_non_nan = next(x for x in wheel if not isnan(x))
first_index = np.where(wheel == first_non_nan)[0]
k = first_index[0]
imp = Imputer(missing_values='NaN', strategy='mean')
wheel = imp.fit_transform(wheel)
wheel = scale_wheel.transform(wheel)
video_pointer.set(1, k)
ret, frame = video_pointer.read()
# crops and converts frame into desired format
frame = cv2.cvtColor(frame[160:400, 100:640], cv2.COLOR_BGR2GRAY)
prvs = frame
nex = frame
# initialize vectors to keep track of data
count = 1
mod = 0
opticals = []
angles = []
frames = []
behavior_data = ep(exp_folder, lims_ID).get_per_frame_data()
labels = [
"chattering", "trunk_present", "grooming", "trunk_absent", "running",
"fidget", "tail_relaxed", "tail_tense", "flailing_present",
"flailing_absent", "walking"
]
Annotators = [
'Mahdi Ramadan', 'Fiona Griffin', 'Nate Berbesque', 'Robert Howard',
'Kyla Mace', 'Twee'
]
# length of movie
limit = int(video_pointer.get(cv2.cv.CV_CAP_PROP_FRAME_COUNT))
fps = video_pointer.get(cv2.cv.CV_CAP_PROP_FPS)
print('creating hf')
# create hdf file
hf = h5py.File(
exp_folder + '\h5 files\\' + 'training_data_' + str(lims_ID) + '.h5',
'w')
while count < limit - first_index[0]:
# get behavior info for frame
behavior = ""
temp = 1
for item in labels:
if behavior_data[temp][k] == 1:
behavior = behavior + " " + item
temp += 1
prvs = nex
frame_data = process_input(prvs)
frames = scale_frame.transform(frame_data)
ret, frame = video_pointer.read()
nex = cv2.cvtColor(frame[160:400, 100:640], cv2.COLOR_BGR2GRAY)
optical = optical_flow(prvs, nex)
opticals = scale_optical.transform(optical['mag'])
angles = scale_angle.transform(optical['ang'])
vector_data = np.concatenate(
(np.reshape(wheel[k], (1)), frames, opticals, angles))
name = 'frame number ' + str(k)
table = hf.create_dataset(name,
data=vector_data,
shape=(1, 5713),
compression=9)
table.attrs['behavior'] = behavior
count += 1
k += 1
if count % 1000 == 0:
print(count)
k += 1
return {'final_data': data, 'y_train': y_train}
def show_frame(frame):
cv2.imshow('image', frame)
cv2.waitKey(0)
cv2.destroyAllWindows()
if __name__ == '__main__':
lims_ID = ['501560436', '501021421', '500860585']
final_data = []
y_train = []
t= 0
for itm in lims_ID:
exl = ep("C:\Users\mahdir\Desktop\Mahdi files", itm)
data = get_data(exl, itm)
if t == 0:
final_data = data['final_data']
y_train= data['y_train']
else:
vector = data['final_data']
final_data = np.vstack((final_data, vector))
y_train = np.concatenate((y_train, data['y_train']))
t += 1
print('feature processing finished')
p = Process(target = run_svm(final_data, y_train), args = (final_data, y_train))
p.start()
p.join()