def evaluate(name, selection_file, sequence_file, outcome_dir, output_dir):
trackers = cvtrack.load_selection(selection_file)
seqs = cvseq.load_selection(sequence_file)
if not os.path.isdir(output_dir):
os.mkdir(output_dir)
output_dir = os.path.join(output_dir, 'plot')
if not os.path.isdir(output_dir):
os.mkdir(output_dir)
if not os.path.isdir(outcome_dir):
os.mkdir(outcome_dir)
m = len(trackers)
n = len(seqs)
outcomes = np.empty((m, n),dtype=np.object)
timings = np.empty((m, n))
timings[:] = np.nan
for i in xrange(m):
tracker = trackers[i]
tracker_outcome_dir = os.path.join(outcome_dir, tracker.name)
if not os.path.exists(tracker_outcome_dir):
os.mkdir(tracker_outcome_dir)
for j in xrange(n):
print('Sequence ' + str(j) + '/' + str(n))
sequence = seqs[j]
if sequence.identifier in tracker.blacklist:
print(sequence.identifier + ' is in blacklist of ' + tracker.name)
continue
sequence_outcome_dir = os.path.join(tracker_outcome_dir, sequence.identifier)
if not os.path.exists(sequence_outcome_dir):
os.mkdir(sequence_outcome_dir)
#Now: Run tracker on sequence or collect existing result
outcome_file = os.path.join(sequence_outcome_dir, 'output.txt')
timing_file = os.path.join(sequence_outcome_dir, 'timing.txt')
if os.path.exists(outcome_file):
print outcome_file, 'already exists, using cached version.'
outcome = np.genfromtxt(outcome_file, delimiter=',')
timing = np.genfromtxt(timing_file, delimiter=',')
else:
try:
[outcome, timing] = tracker.run(sequence)
#Cache output for next run
np.savetxt(outcome_file, outcome, delimiter=',')
with open(timing_file, 'w') as f:
f.write('{0}\n'.format(timing))
except Exception:
print('Tracker ' + tracker.name + ' failed on sequence ' + sequence.identifier)
outcome = None
timing = None
if outcome is not None:
nPoints = outcome.shape[1] / 2.0
if not nPoints == 2:
#Convert polygon to bounding box
outcome = cvbb.poly2bb(outcome)
outcomes[i,j] = outcome
timings[i,j] = timing
#Then pass result to evaluation function
cvpr2013(name, trackers, seqs, outcomes, output_dir)
wacv2014(name, trackers, seqs, outcomes, output_dir)
table(name, trackers, seqs, outcomes, output_dir)
compute_timing(name, trackers, seqs, timings, output_dir)
with open(os.path.join(output_dir, name + '_list.txt'), 'w') as f:
for tracker in trackers:
f.write(tracker.name + '\n')
def evaluate(name, selection_file, sequence_file, outcome_dir, output_dir):
trackers = cvtrack.load_selection(selection_file)
seqs = cvseq.load_selection(sequence_file)
output_dir = os.path.join(output_dir, 'plot')
m = len(trackers)
n = len(seqs)
outcomes = np.empty((m, n), dtype=np.object)
timings = np.empty((m, n))
timings[:] = np.nan
for i in xrange(m):
tracker = trackers[i]
tracker_outcome_dir = os.path.join(outcome_dir, tracker.name)
if not os.path.exists(tracker_outcome_dir):
os.mkdir(tracker_outcome_dir)
for j in xrange(n):
print('Sequence ' + str(j) + '/' + str(n))
sequence = seqs[j]
if sequence.identifier in tracker.blacklist:
print(sequence.identifier + ' is in blacklist of ' +
tracker.name)
continue
sequence_outcome_dir = os.path.join(tracker_outcome_dir,
sequence.identifier)
if not os.path.exists(sequence_outcome_dir):
os.mkdir(sequence_outcome_dir)
#Now: Run tracker on sequence or collect existing result
outcome_file = os.path.join(sequence_outcome_dir, 'output.txt')
timing_file = os.path.join(sequence_outcome_dir, 'timing.txt')
if os.path.exists(outcome_file):
print outcome_file, 'already exists, using cached version.'
outcome = np.genfromtxt(outcome_file, delimiter=',')
timing = np.genfromtxt(timing_file, delimiter=',')
else:
try:
[outcome, timing] = tracker.run(sequence)
#Cache output for next run
np.savetxt(outcome_file, outcome, delimiter=',')
with open(timing_file, 'w') as f:
f.write('{0}\n'.format(timing))
except Exception:
print('Tracker ' + tracker.name + ' failed on sequence ' +
sequence.identifier)
outcome = None
timing = None
if outcome is not None:
nPoints = outcome.shape[1] / 2.0
if not nPoints == 2:
#Convert polygon to bounding box
outcome = cvbb.poly2bb(outcome)
outcomes[i, j] = outcome
timings[i, j] = timing
#Then pass result to evaluation function
cvpr2013(name, trackers, seqs, outcomes, output_dir)
wacv2014(name, trackers, seqs, outcomes, output_dir)
table(name, trackers, seqs, outcomes, output_dir)
compute_timing(name, trackers, seqs, timings, output_dir)
with open(os.path.join(output_dir, name + '_list.txt'), 'w') as f:
for tracker in trackers:
f.write(tracker.name + '\n')
def prop_annotate(seq, prop):
#Create path to annotation folder
annot_folder = os.path.join('annotations', seq.identifier)
#Create folder if it doesn't exist
if not os.path.exists(annot_folder):
os.mkdir(annot_folder)
annotation_file = os.path.join(annot_folder, prop + '.label')
if os.path.exists(annotation_file):
annot = np.genfromtxt(annotation_file).astype(np.bool)
if annot.shape[0] != seq.num_frames:
raise Exception("Number of frames in annotation file differs from number of frames in sequence.")
else:
annot = np.zeros((seq.num_frames,1), dtype = np.bool)
frame = 0
start_frame = None
gt = cvbb.poly2bb(seq.gt)
while True:
#Read image
im_path = seq.im_list[frame]
if not os.path.exists(im_path):
raise Exception(im_path + ' does not exist')
im = cv2.imread(im_path)
#Display annotations
im_text = str(frame+1)
if annot[frame] == True:
im_text += ' ' + prop
# #Draw groundtruth
gt_frame = gt[frame]
if not np.any(np.isnan(gt_frame)):
tl = cvutil.array_to_int_tuple(gt_frame[:2])
br = cvutil.array_to_int_tuple(gt_frame[:2] + gt_frame[2:4])
cv2.rectangle(im, tl, br, (255,0,0))
cv2.putText(im, im_text, (10,30), cv2.FONT_HERSHEY_SIMPLEX, 0.5, 255)
#Show image, wait for key
cv2.imshow(seq.identifier, im)
key = cv2.waitKey(0)
key = chr(key & 255)
#Handle key
if key == 's':
print 'Start frame of annotation set.'
start_frame = frame
if key == 'e':
if start_frame is not None:
end_frame = frame+1
annot[start_frame:end_frame] = True
start_frame = None
print 'End frame of annotation set.'
else:
print 'No start frame set for end frame'
if key == 'd':
print 'Deleting continuous annotation'
if annot[frame]:
#Delete forward
frame_forward = frame
while frame_forward < seq.num_frames and annot[frame_forward] == True:
annot[frame_forward] = False
frame_forward += 1
frame_backward = frame-1
while frame_backward >= 0 and annot[frame_backward] == True:
annot[frame_backward] = False
frame_backward -= 1
if key == 'x':
print 'Deleting single annotation'
annot[frame] = False
#Standard movement
if key == 'S' or key == ' ':
frame += 1
if key == 'Q':
frame -= 1 #Go back one frame
if key == 'P':
frame = 0
if key == 'W':
frame = seq.num_frames-1
if key == 'm':
frame = seq.num_frames / 2
if key == 'V':
frame += seq.num_frames / 100
if key == 'U':
frame -= seq.num_frames / 100
if key == 'q':
break
#Don't go out of bounds
frame = max(frame, 0)
frame = min(frame, seq.num_frames-1)
#When loop was exited, save file
annotation_file = os.path.join(annot_folder, prop + '.label')
np.savetxt(annotation_file, annot, fmt = '%d')