def plot_trajectory_on_background(self, bagfile):
if not os.path.exists(self._tempdir + 'targets.png'):
self.plot_targets_on_background()
im = cv2.imread(self._tempdir + 'targets.png')
positions = utilities.get_positions_from_bag(bagfile)
distances = self.get_dist_to_nearest_target(bagfile)
for ts in positions.index:
x = int(np.rint(positions.fly_x[ts]))
y = int(np.rint(positions.fly_y[ts]))
d = int(np.rint(distances.dtarget[ts]))
fraction = (float(ts)/len(positions))
if fraction <= 0.5:
red = 255.0*2*(fraction)
green = 255.0*2*(fraction)
blue = 255.0
if fraction > 0.5:
red = 255.0
green = 255.0 - 255.0*2*(fraction - 0.5)
blue = 255.0 - 255.0*2*(fraction - 0.5)
cv2.circle(im, (x,y), 1, (blue, green, red), -1)
cv2.imshow('traj', im)
cv2.imwrite((self._tempdir + 'trajectory.png'), im)
cv2.destroyAllWindows()
def plot_trajectory_on_background(self, bagfile):
if not os.path.exists(self._tempdir + 'targets.png'):
self.plot_targets_on_background()
im = cv2.imread(self._tempdir + 'targets.png')
positions = utilities.get_positions_from_bag(bagfile)
distances = self.get_dist_to_nearest_target(bagfile)
for ts in positions.index:
x = int(np.rint(positions.fly_x[ts]))
y = int(np.rint(positions.fly_y[ts]))
d = int(np.rint(distances.dtarget[ts]))
fraction = (float(ts) / len(positions))
if fraction <= 0.5:
red = 255.0 * 2 * (fraction)
green = 255.0 * 2 * (fraction)
blue = 255.0
if fraction > 0.5:
red = 255.0
green = 255.0 - 255.0 * 2 * (fraction - 0.5)
blue = 255.0 - 255.0 * 2 * (fraction - 0.5)
cv2.circle(im, (x, y), 1, (blue, green, red), -1)
cv2.imshow('traj', im)
cv2.imwrite((self._tempdir + 'trajectory.png'), im)
cv2.destroyAllWindows()
def get_dist_to_nearest_target(self, bagfile):
self._targets = self.detect_targets()
positions = utilities.get_positions_from_bag(bagfile)
distances = DataFrame()
for target in range(len(self._targets)):
px, py, pr = self._targets[target]
distances['d'+ str(target)] = ((((positions['fly_x'] - px)/5.2)**2 + ((positions['fly_y'] - py)/4.8)**2)**0.5)# - pr
distances['Timestamp'] = positions.Timestamp
distances = utilities.convert_timestamps(distances)
self.dtarget = DataFrame(distances.min(axis=1), columns=['dtarget'])
return self.dtarget
def get_dist_to_nearest_target(self, bagfile):
self._targets = self.detect_targets()
positions = utilities.get_positions_from_bag(bagfile)
distances = DataFrame()
for target in range(len(self._targets)):
px, py, pr = self._targets[target]
distances['d' + str(target)] = (
(((positions['fly_x'] - px) / 5.2)**2 +
((positions['fly_y'] - py) / 4.8)**2)**0.5) # - pr
distances['Timestamp'] = positions.Timestamp
distances = utilities.convert_timestamps(distances)
self.dtarget = DataFrame(distances.min(axis=1), columns=['dtarget'])
return self.dtarget
def sync_jaaba_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
JAABA_CSV = FMF_DIR + '/registered_trx.csv'
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, JAABA)
jaaba_data = pd.read_csv(JAABA_CSV, sep=',', names=['Timestamp','Length','Width','Theta','Left','Right'], index_col=False)
jaaba_data[['Length','Width','Left','Right']] = jaaba_data[['Length','Width','Left','Right']].astype(np.float64)
jaaba_data = convert_timestamps(jaaba_data)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
jaaba_data['Laser0_state'] = laser_states['Laser0_state'].asof(jaaba_data.index).fillna(value=1)
jaaba_data['Laser1_state'] = laser_states['Laser1_state'].asof(jaaba_data.index).fillna(value=0) #YAY!
jaaba_data['Laser2_state'] = laser_states['Laser2_state'].asof(jaaba_data.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
jaaba_data['Laser0_state'] = 0
jaaba_data['Laser2_state'] = 0
jaaba_data['Laser1_state'] = 0
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
jaaba_data['fly_x'] = positions['fly_x'].asof(jaaba_data.index).fillna(value=0)
jaaba_data['fly_y'] = positions['fly_y'].asof(jaaba_data.index).fillna(value=0)
jaaba_data['dtarget'] = targets.get_dist_to_nearest_target(BAG_FILE)['dtarget'].asof(jaaba_data.index).fillna(value=0)
jaaba_data['Timestamp'] = jaaba_data.index #silly pandas bug for subtracting from datetimeindex...
try:
jaaba_data['synced_time'] = jaaba_data['Timestamp'] - jaaba_data.Timestamp[(jaaba_data.Laser2_state + jaaba_data.Laser1_state) > 0.001].index[0]
except:
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame0. "
jaaba_data['synced_time'] = jaaba_data['Timestamp'] - jaaba_data.Timestamp.index[0]
jaaba_data.index = jaaba_data.synced_time
jaaba_data.index = pd.to_datetime(jaaba_data.index)
### WING EXTENSION ###
jaaba_data['maxWingAngle'] = get_absmax(jaaba_data[['Left','Right']])
#jaaba_data[jaaba_data['maxWingAngle'] > 3.1] = np.nan
program = None
if program == 'IRR':
BEGINNING =jaaba_data.Timestamp[jaaba_data.synced_time >= -30000000000].index[0]#jaaba_data.Timestamp.index[0]
#FIRST_IR_ON = jaaba_data.Timestamp[((jaaba_data.Laser1_state > 0.001) & (jaaba_data.synced_time >= -1))].index[0]
FIRST_IR_ON = jaaba_data.Timestamp[jaaba_data.synced_time >= 0].index[0]
#FIRST_IR_OFF = jaaba_data.Timestamp[((jaaba_data.Laser1_state > 0.001) & (jaaba_data.synced_time <= 120))].index[-1]
FIRST_IR_OFF = jaaba_data.Timestamp[jaaba_data.synced_time >= 60000000000].index[0]
RED_ON = jaaba_data.Timestamp[jaaba_data.Laser2_state > 0.001].index[0]
RED_OFF = jaaba_data.Timestamp[jaaba_data.Laser2_state > 0.001].index[-1]
SECOND_IR_ON = jaaba_data.Timestamp[jaaba_data.synced_time >=320000000000].index[0]
#SECOND_IR_ON = jaaba_data.Timestamp[((jaaba_data.Laser1_state > 0.001) & (jaaba_data.synced_time >= 120))].index[0]
SECOND_IR_OFF = jaaba_data.Timestamp[jaaba_data.Laser1_state > 0.001].index[-1]
END = jaaba_data.Timestamp.index[-1]
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, BEGINNING, FIRST_IR_ON, '1-prestim', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, FIRST_IR_ON, FIRST_IR_OFF, '2-IR1', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, FIRST_IR_OFF, RED_ON, '3-post-IR1', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, RED_ON,RED_OFF, '4-red', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE,RED_OFF, SECOND_IR_ON,'5-post-red', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE,SECOND_IR_ON,SECOND_IR_OFF,'6-IR2', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE,SECOND_IR_OFF,END,'7-post-IR2', background=False)
if program == 'dark':
BEGINNING =jaaba_data.Timestamp[jaaba_data.synced_time >= -30000000000].index[0]
print set(jaaba_data.Laser0_state), set(jaaba_data.Laser1_state), set(jaaba_data.Laser2_state)
STIM_ON = jaaba_data.Timestamp[jaaba_data.Laser1_state > 0.001].index[0]
STIM_OFF = jaaba_data.Timestamp[jaaba_data.Laser1_state > 0.001].index[-1]
LIGHTS_OUT = jaaba_data.Timestamp[jaaba_data.Laser0_state < 0.001].index[0]
LIGHTS_ON = jaaba_data.Timestamp[jaaba_data.Laser0_state < 0.001].index[-1]
END = jaaba_data.Timestamp.index[-1]
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, BEGINNING, STIM_ON, '1-prestim', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, STIM_ON,STIM_OFF, '2-stim', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, STIM_OFF, LIGHTS_OUT,'3-post-stim', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, LIGHTS_OUT, LIGHTS_ON,'4-DARK', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, LIGHTS_ON, END,'7-light', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE)
### ABDOMINAL BENDING ###
jaaba_data[jaaba_data['Length'] > 110] = np.nan #discard frames with bogus length. *************
jaaba_data[jaaba_data['Length'] < 60] = np.nan #discard frames with bogus length.
trace = plot_single_trace(jaaba_data)
trace.savefig(JAABA + 'TRACES/' + FLY_ID + '.png')
plt.close('all')
###FIRST COURTSHIP BOUT AFTER STIMULUS###
#courted, latency = latency_measures(jaaba_data)
if 'binsize' in globals():
jaaba_data = bin_data(jaaba_data, binsize)
jaaba_data.to_pickle(JAABA + 'JAR/' + FLY_ID + '_' + binsize + '_fly.pickle')
else:
return jaaba_data, courted, latency
def sync_jaaba_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
JAABA_CSV = FMF_DIR + '/registered_trx.csv'
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, JAABA)
jaaba_data = pd.read_csv(JAABA_CSV, sep=',', names=['Timestamp','Length','Width','Theta','Left','Right'], index_col=False)
jaaba_data[['Length','Width','Left','Right']] = jaaba_data[['Length','Width','Left','Right']].astype(np.float64)
jaaba_data = convert_timestamps(jaaba_data)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
jaaba_data['Laser0_state'] = laser_states['Laser0_state'].asof(jaaba_data.index).fillna(value=1)
jaaba_data['Laser1_state'] = laser_states['Laser1_state'].asof(jaaba_data.index).fillna(value=0) #YAY!
jaaba_data['Laser2_state'] = laser_states['Laser2_state'].asof(jaaba_data.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
jaaba_data['Laser0_state'] = 0
jaaba_data['Laser2_state'] = 0
jaaba_data['Laser1_state'] = 0
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
jaaba_data['fly_x'] = positions['fly_x'].asof(jaaba_data.index).fillna(value=0)
jaaba_data['fly_y'] = positions['fly_y'].asof(jaaba_data.index).fillna(value=0)
jaaba_data['dtarget'] = targets.get_dist_to_nearest_target(BAG_FILE)['dtarget'].asof(jaaba_data.index).fillna(value=0)
jaaba_data['Timestamp'] = jaaba_data.index #silly pandas bug for subtracting from datetimeindex...
number_of_bouts, bout_duration, first_TS, last_TS = utilities.detect_stim_bouts(jaaba_data, 'Laser1_state')
try:
jaaba_data['synced_time'] = jaaba_data['Timestamp'] - last_TS
except:
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame 1500. "
jaaba_data['synced_time'] = jaaba_data['Timestamp'] - jaaba_data.Timestamp.index[1500]
jaaba_data.index = jaaba_data.synced_time
jaaba_data.index = pd.to_datetime(jaaba_data.index)
### WING EXTENSION ###
jaaba_data['maxWingAngle'] = get_absmax(jaaba_data[['Left','Right']])
#jaaba_data[jaaba_data['maxWingAngle'] > 3.1] = np.nan
number_of_bouts, stim_duration, first_TS, last_TS = utilities.detect_stim_bouts(jaaba_data, 'Laser2_state') #HACK DANNO
jaaba_data['stim_duration'] = stim_duration
BEGINNING =jaaba_data.Timestamp.index[0]
END = jaaba_data.Timestamp.index[-1]
print BEGINNING
print first_TS
print last_TS
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, BEGINNING, first_TS, '1-prestim', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, first_TS,last_TS, '2-red', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, last_TS, END,'3-post-red', background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE)
### ABDOMINAL BENDING ###
jaaba_data[jaaba_data['Length'] > 110] = np.nan #discard frames with bogus length. *************
jaaba_data[jaaba_data['Length'] < 60] = np.nan #discard frames with bogus length.
trace = plot_single_trace(jaaba_data)
trace.savefig(JAABA + 'TRACES/' + FLY_ID + '.png')
plt.close('all')
if 'binsize' in globals():
jaaba_data = bin_data(jaaba_data, binsize)
jaaba_data.to_pickle(JAABA + 'JAR/' + FLY_ID + '_' + binsize + '_fly.pickle')
else:
return jaaba_data, courted, latency
def sync_video_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, MAIN_DIRECTORY+"/")
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
dtarget = targets.get_dist_to_nearest_target(BAG_FILE)['dtarget']
arenaCtr, Arena_radius = targets._arena.circ
for x in utilities.find_files(FMF_DIR, '*zoom*.fmf'):
ZOOM_FMF = x
if not os.path.exists(MAIN_DIRECTORY + '/JAR/wing_data') ==True:
os.makedirs(MAIN_DIRECTORY + '/JAR/wing_data')
if args.save_wingext_pngs == True:
if not os.path.exists(FMF_DIR + '/TRACKING_RESULTS'):
os.makedirs(FMF_DIR + '/TRACKING_RESULTS')
wingImageDir = FMF_DIR + '/TRACKING_RESULTS'
else:
wingImageDir=None
if not os.path.exists(MAIN_DIRECTORY + '/JAR/wing_data/'+ FLY_ID + '_wing_angles.pickle') ==True:
wings = wing_detector.WingDetector(ZOOM_FMF, BAG_FILE, dtarget, arenaCtr, wingImageDir )
wings.execute()
wings.wingData.to_pickle(MAIN_DIRECTORY + '/JAR/wing_data/'+ FLY_ID + '_wing_angles.pickle')
videoData = pd.read_pickle(MAIN_DIRECTORY + '/JAR/wing_data/'+ FLY_ID + '_wing_angles.pickle')
videoData[['Length','Width','Left','Right']] = videoData[['Length','Width','Left','Right']].astype(np.float64)
videoData = convert_timestamps(videoData)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
videoData['Laser0_state'] = laser_states['Laser0_state'].asof(videoData.index).fillna(value=1)
videoData['Laser1_state'] = laser_states['Laser1_state'].asof(videoData.index).fillna(value=0) #YAY!
videoData['Laser2_state'] = laser_states['Laser2_state'].asof(videoData.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
videoData['Laser0_state'] = 0
videoData['Laser2_state'] = 0
videoData['Laser1_state'] = 0
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
videoData['fly_x'] = positions['fly_x'].asof(videoData.index).fillna(value=0)
videoData['fly_y'] = positions['fly_y'].asof(videoData.index).fillna(value=0)
videoData['dtarget'] = dTarget.asof(videoData.index).fillna(value=0)
videoData['Timestamp'] = videoData.index #silly pandas bug for subtracting from datetimeindex...
try:
videoData['synced_time'] = videoData['Timestamp'] - videoData.Timestamp[(videoData.Laser2_state + videoData.Laser1_state) > 0.001].index[0]
except:
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame0. "
videoData['synced_time'] = videoData['Timestamp'] - videoData.Timestamp.index[0]
videoData.index = videoData.synced_time
videoData.index = pd.to_datetime(videoData.index)
### WING EXTENSION ###
videoData['maxWingAngle'] = get_absmax(videoData[['Left','Right']])
#videoData[videoData['maxWingAngle'] > 3.1] = np.nan
program = 'dark' #FIXME
if program == 'IRR':
BEGINNING =videoData.Timestamp[videoData.synced_time >= -30000000000].index[0]#videoData.Timestamp.index[0]
#FIRST_IR_ON = videoData.Timestamp[((videoData.Laser1_state > 0.001) & (videoData.synced_time >= -1))].index[0]
FIRST_IR_ON = videoData.Timestamp[videoData.synced_time >= 0].index[0]
#FIRST_IR_OFF = videoData.Timestamp[((videoData.Laser1_state > 0.001) & (videoData.synced_time <= 120))].index[-1]
FIRST_IR_OFF = videoData.Timestamp[videoData.synced_time >= 60000000000].index[0]
RED_ON = videoData.Timestamp[videoData.Laser2_state > 0.001].index[0]
RED_OFF = videoData.Timestamp[videoData.Laser2_state > 0.001].index[-1]
SECOND_IR_ON = videoData.Timestamp[videoData.synced_time >=320000000000].index[0]
#SECOND_IR_ON = videoData.Timestamp[((videoData.Laser1_state > 0.001) & (videoData.synced_time >= 120))].index[0]
SECOND_IR_OFF = videoData.Timestamp[videoData.Laser1_state > 0.001].index[-1]
END = videoData.Timestamp.index[-1]
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, BEGINNING, FIRST_IR_ON, '1-prestim', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, FIRST_IR_ON, FIRST_IR_OFF, '2-IR1', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, FIRST_IR_OFF, RED_ON, '3-post-IR1', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, RED_ON,RED_OFF, '4-red', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE,RED_OFF, SECOND_IR_ON,'5-post-red', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE,SECOND_IR_ON,SECOND_IR_OFF,'6-IR2', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE,SECOND_IR_OFF,END,'7-post-IR2', background=False)
if program == 'dark':
BEGINNING =videoData.Timestamp[videoData.synced_time >= -30000000000].index[0]
print set(videoData.Laser0_state), set(videoData.Laser1_state), set(videoData.Laser2_state)
STIM_ON = videoData.Timestamp[videoData.Laser1_state > 0.001].index[0]
STIM_OFF = videoData.Timestamp[videoData.Laser1_state > 0.001].index[-1]
LIGHTS_OUT = videoData.Timestamp[videoData.Laser0_state < 0.001].index[0]
LIGHTS_ON = videoData.Timestamp[videoData.Laser0_state < 0.001].index[-1]
END = videoData.Timestamp.index[-1]
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, BEGINNING, STIM_ON, '1-prestim', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, STIM_ON,STIM_OFF, '2-stim', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, STIM_OFF, LIGHTS_OUT,'3-post-stim', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, LIGHTS_OUT, LIGHTS_ON,'4-DARK', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE, LIGHTS_ON, END,'7-light', background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE)
### ABDOMINAL BENDING ###
videoData[videoData['Length'] > 110] = np.nan #discard frames with bogus length. *************
videoData[videoData['Length'] < 60] = np.nan #discard frames with bogus length.
trace = plot_single_trace(videoData)
trace.savefig(MAIN_DIRECTORY + '/TRACES/' + FLY_ID + '.png')
plt.close('all')
###FIRST COURTSHIP BOUT AFTER STIMULUS###
#courted, latency = latency_measures(videoData)
if 'binsize' in globals():
videoData = bin_data(videoData, binsize)
videoData.to_pickle(MAIN_DIRECTORY + '/JAR/' + FLY_ID + '_' + binsize + '_fly.pickle')
else:
return videoData, courted, latency
def sync_jaaba_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
DATADIR_CSV = FMF_DIR + '/registered_trx.csv'
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, DATADIR)
FLY_IDx = FMF_DIR.split('/')[-1]
EXP_ID, DATE, TIME = FLY_IDx.split('_', 4)[0:3]
for x in glob.glob(FMF_DIR +'/*zoom*'+'*.fmf'):
ZOOM_FMF = x
if not os.path.exists(FMF_DIR + '/TRACKING_FRAMES') == True:
os.makedirs(FMF_DIR + '/TRACKING_FRAMES')
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
if (os.path.exists(FMF_DIR + '/frame_by_frame_synced.pickle')) and not RETRACK:
datadf = pd.read_pickle(FMF_DIR + '/frame_by_frame_synced.pickle')
exp_group = EXP_ID #QUICKFIX. FIXME
else:
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
dtarget = targets.get_dist_to_nearest_target(BAG_FILE)['dtarget']
(arena_centre), arena_radius = targets._arena.circ
if MAKE_MOVIES:
TRACKING_DIRECTORY = FMF_DIR + '/TRACKING_FRAMES/'
else:
TRACKING_DIRECTORY = None
if (os.path.exists(FMF_DIR + '/wingdata.pickle')) and not RETRACK:
datadf = pd.read_pickle(FMF_DIR + '/wingdata.pickle')
datadf.columns= ['BodyAxis','leftAngle','leftWingLength','Length','rightAngle','rightWingLength','target_angle_TTM',
'target_distance_TTM','Timestamp','Width']
else:
try:
wings = wing_detector.WingDetector(ZOOM_FMF, BAG_FILE, dtarget, arena_centre, RETRACK, TRACKING_DIRECTORY )
wings.execute()
wings.wingData.columns= ['BodyAxis','leftAngle','leftWingLength','Length','rightAngle','rightWingLength','target_angle_TTM',
'target_distance_TTM','Timestamp','Width']
wings.wingData.to_pickle(FMF_DIR + '/wingdata.pickle')
wings.tracking_info.to_pickle(FMF_DIR + '/tracking_info.pickle')
datadf = DataFrame(wings.wingData)
if MAKE_MOVIES:
utilities.call_command("ffmpeg -f image2 -r 15 -i "+ TRACKING_DIRECTORY + "_tmp%05d.png -vf scale=iw/2:-1 -vcodec mpeg4 -b 8000k -y " + FMF_DIR + "/tracked_movie.mp4;")
utilities.call_command("rm -r " + TRACKING_DIRECTORY)
#wings.make_movie(wings._tempdir, wings.fmf_file.rsplit('/',1)[0]+'/tracked_movie.mp4',15)
except Exception,e:
traceback.print_exc()
print 'ERROR PROCESSING WING TRACKING...', FMF_DIR
print str(e)
return
datadf['Frame_number'] = datadf.index
datadf = convert_timestamps(datadf)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
datadf['Laser0_state'] = laser_states['Laser0_state'].asof(datadf.index).fillna(value=1)
datadf['Laser1_state'] = laser_states['Laser1_state'].asof(datadf.index).fillna(value=0) #YAY!
datadf['Laser2_state'] = laser_states['Laser2_state'].asof(datadf.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
datadf['Laser0_state'] = 0
datadf['Laser2_state'] = 0
datadf['Laser1_state'] = 0
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
datadf['fly_x'] = positions['fly_x'].asof(datadf.index).fillna(method='pad')
datadf['fly_y'] = positions['fly_y'].asof(datadf.index).fillna(method='pad')
datadf['dcentre'] = np.sqrt(((datadf['fly_x']-arena_centre[0])/4.8)**2 + ((datadf['fly_y']-arena_centre[1])/5.2)**2)
datadf['dtarget_temp'] = targets.get_dist_to_nearest_target(BAG_FILE)['dtarget'].asof(datadf.index).fillna(method='pad')
datadf.loc[datadf['dtarget_temp'] >=6.0, 'target_distance_TTM'] = np.nan
datadf['dtarget'] = datadf['target_distance_TTM'].fillna(datadf['dtarget_temp'])
datadf['Timestamp'] = datadf.index #silly pandas bug for subtracting from datetimeindex...
#number_of_bouts, bout_duration, first_TS, last_TS = utilities.detect_stim_bouts(datadf, 'Laser1_state')
number_of_bouts, stim_duration, first_TS, last_TS = utilities.detect_stim_bouts(datadf, 'Laser2_state') #HACK DANNO
datadf['stim_duration'] = stim_duration
try:
if stim_duration >= 0:
datadf['synced_time'] = datadf['Timestamp'] - datadf.Timestamp[(datadf.Laser2_state + datadf.Laser1_state) > 0.001].index[0]
#elif stim_duration == 0:
# datadf['synced_time'] = datadf['Timestamp'] - datadf.Timestamp.index[0] - pd.to_datetime('60s')
except Exception,e:
traceback.print_exc()
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame 900. "
datadf['synced_time'] = datadf['Timestamp'] - datadf.Timestamp.index[900]
print str(e)
def sync_jaaba_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
JAABA_CSV = FMF_DIR + '/registered_trx.csv'
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, JAABA)
jaaba_data = pd.read_csv(
JAABA_CSV,
sep=',',
names=['Timestamp', 'Length', 'Width', 'Theta', 'Left', 'Right'],
index_col=False)
jaaba_data[['Length', 'Width', 'Left',
'Right']] = jaaba_data[['Length', 'Width', 'Left',
'Right']].astype(np.float64)
jaaba_data = convert_timestamps(jaaba_data)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
jaaba_data['Laser0_state'] = laser_states['Laser0_state'].asof(
jaaba_data.index).fillna(value=0)
jaaba_data['Laser1_state'] = laser_states['Laser1_state'].asof(
jaaba_data.index).fillna(value=0) #YAY!
jaaba_data['Laser2_state'] = laser_states['Laser2_state'].asof(
jaaba_data.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
jaaba_data['Laser0_state'] = 0
jaaba_data['Laser2_state'] = 0
jaaba_data['Laser1_state'] = 0
print '\t acquired stimulus data...'
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
jaaba_data['fly_x'] = positions['fly_x'].asof(
jaaba_data.index).fillna(value=0)
jaaba_data['fly_y'] = positions['fly_y'].asof(
jaaba_data.index).fillna(value=0)
jaaba_data['dtarget'] = targets.get_dist_to_nearest_target(
BAG_FILE)['dtarget'].asof(jaaba_data.index).fillna(value=0)
print '\t computed distance to target...'
### WING EXTENSION ###
jaaba_data['maxWingAngle'] = get_absmax(jaaba_data[['Left', 'Right']])
jaaba_data[jaaba_data['maxWingAngle'] > 2.1] = np.nan
print '\t computed maximum wing angle...'
### PLOT TRAJECTORIES DURING EXPERIMENT SECTIONS ###
first_IR = jaaba_data[(jaaba_data.Laser1_state) == 1].index[0]
last_IR = jaaba_data[(jaaba_data.Laser1_state) == 1].index[-1]
#first_RED = jaaba_data.Timestamp[(jaaba_data.Laser2_state) == 1].index[0]
#last_RED = jaaba_data.Timestamp[(jaaba_data.Laser2_state) == 1].index[-1]
first_L0 = jaaba_data[(jaaba_data.Laser0_state) == 0].index[0]
last_L0 = jaaba_data[(jaaba_data.Laser0_state) == 0].index[-1]
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE,
jaaba_data.index[0], first_IR,
'1_pre-stim')
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, first_IR,
last_IR, '2_stim')
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, last_IR,
first_L0, '3_post-stim')
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, first_L0,
last_L0, '4_dark')
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, last_L0,
jaaba_data.index[-1], '5_post-dark')
print '\t generated wing extension 2D trajectory plots...'
### ABDOMINAL BENDING ###
jaaba_data[jaaba_data['Length'] >
110] = np.nan #discard frames with bogus length. *************
jaaba_data[
jaaba_data['Length'] < 60] = np.nan #discard frames with bogus length.
#SYNCHRONIZE BY FIRST STIMULUS TIME
jaaba_data[
'Timestamp'] = jaaba_data.index #silly pandas bug for subtracting from datetimeindex...
try:
jaaba_data[
'synced_time'] = jaaba_data['Timestamp'] - jaaba_data.Timestamp[
(jaaba_data.Laser2_state +
jaaba_data.Laser1_state) > 0.001].index[0]
except:
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame0. "
jaaba_data['synced_time'] = jaaba_data[
'Timestamp'] - jaaba_data.Timestamp.index[0]
jaaba_data.index = jaaba_data.synced_time
jaaba_data.index = pd.to_datetime(jaaba_data.index)
trace = plot_single_trace(jaaba_data)
trace.savefig(JAABA + 'TRACES/' + FLY_ID + '.png')
plt.close('all')
###FIRST COURTSHIP BOUT AFTER STIMULUS###
#courted, latency = latency_measures(jaaba_data)
if 'binsize' in globals():
jaaba_data = bin_data(jaaba_data, binsize)
jaaba_data.to_pickle(JAABA + 'JAR/' + FLY_ID + '_' + binsize +
'_fly.pickle')
else:
return jaaba_data, courted, latency
def sync_jaaba_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
JAABA_CSV = FMF_DIR + '/registered_trx.csv'
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, JAABA)
jaaba_data = pd.read_csv(JAABA_CSV, sep=',', names=['Timestamp','Length','Width','Theta','Left','Right'], index_col=False)
jaaba_data[['Length','Width','Left','Right']] = jaaba_data[['Length','Width','Left','Right']].astype(np.float64)
jaaba_data = convert_timestamps(jaaba_data)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
jaaba_data['Laser0_state'] = laser_states['Laser0_state'].asof(jaaba_data.index).fillna(value=0)
jaaba_data['Laser1_state'] = laser_states['Laser1_state'].asof(jaaba_data.index).fillna(value=0) #YAY!
jaaba_data['Laser2_state'] = laser_states['Laser2_state'].asof(jaaba_data.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
jaaba_data['Laser0_state'] = 0
jaaba_data['Laser2_state'] = 0
jaaba_data['Laser1_state'] = 0
print '\t acquired stimulus data...'
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
jaaba_data['fly_x'] = positions['fly_x'].asof(jaaba_data.index).fillna(value=0)
jaaba_data['fly_y'] = positions['fly_y'].asof(jaaba_data.index).fillna(value=0)
jaaba_data['dtarget'] = targets.get_dist_to_nearest_target(BAG_FILE)['dtarget'].asof(jaaba_data.index).fillna(value=0)
print '\t computed distance to target...'
### WING EXTENSION ###
jaaba_data['maxWingAngle'] = get_absmax(jaaba_data[['Left','Right']])
jaaba_data[jaaba_data['maxWingAngle'] > 2.1] = np.nan
print '\t computed maximum wing angle...'
### PLOT TRAJECTORIES DURING EXPERIMENT SECTIONS ###
first_IR = jaaba_data[(jaaba_data.Laser1_state) == 1].index[0]
last_IR = jaaba_data[(jaaba_data.Laser1_state) == 1].index[-1]
#first_RED = jaaba_data.Timestamp[(jaaba_data.Laser2_state) == 1].index[0]
#last_RED = jaaba_data.Timestamp[(jaaba_data.Laser2_state) == 1].index[-1]
first_L0 = jaaba_data[(jaaba_data.Laser0_state) == 0].index[0]
last_L0 = jaaba_data[(jaaba_data.Laser0_state) == 0].index[-1]
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, jaaba_data.index[0], first_IR, '1_pre-stim')
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, first_IR, last_IR, '2_stim')
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, last_IR, first_L0, '3_post-stim')
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, first_L0, last_L0, '4_dark')
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, last_L0, jaaba_data.index[-1], '5_post-dark')
print '\t generated wing extension 2D trajectory plots...'
### ABDOMINAL BENDING ###
jaaba_data[jaaba_data['Length'] > 110] = np.nan #discard frames with bogus length. *************
jaaba_data[jaaba_data['Length'] < 60] = np.nan #discard frames with bogus length.
#SYNCHRONIZE BY FIRST STIMULUS TIME
jaaba_data['Timestamp'] = jaaba_data.index #silly pandas bug for subtracting from datetimeindex...
try:
jaaba_data['synced_time'] = jaaba_data['Timestamp'] - jaaba_data.Timestamp[(jaaba_data.Laser2_state + jaaba_data.Laser1_state) > 0.001].index[0]
except:
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame0. "
jaaba_data['synced_time'] = jaaba_data['Timestamp'] - jaaba_data.Timestamp.index[0]
jaaba_data.index = jaaba_data.synced_time
jaaba_data.index = pd.to_datetime(jaaba_data.index)
trace = plot_single_trace(jaaba_data)
trace.savefig(JAABA + 'TRACES/' + FLY_ID + '.png')
plt.close('all')
###FIRST COURTSHIP BOUT AFTER STIMULUS###
#courted, latency = latency_measures(jaaba_data)
if 'binsize' in globals():
jaaba_data = bin_data(jaaba_data, binsize)
jaaba_data.to_pickle(JAABA + 'JAR/' + FLY_ID + '_' + binsize + '_fly.pickle')
else:
return jaaba_data, courted, latency
def sync_video_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, MAIN_DIRECTORY + "/")
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
dtarget = targets.get_dist_to_nearest_target(BAG_FILE)['dtarget']
arenaCtr, Arena_radius = targets._arena.circ
for x in utilities.find_files(FMF_DIR, '*zoom*.fmf'):
ZOOM_FMF = x
if not os.path.exists(MAIN_DIRECTORY + '/JAR/wing_data') == True:
os.makedirs(MAIN_DIRECTORY + '/JAR/wing_data')
if args.save_wingext_pngs == True:
if not os.path.exists(FMF_DIR + '/TRACKING_RESULTS'):
os.makedirs(FMF_DIR + '/TRACKING_RESULTS')
wingImageDir = FMF_DIR + '/TRACKING_RESULTS'
else:
wingImageDir = None
if not os.path.exists(MAIN_DIRECTORY + '/JAR/wing_data/' + FLY_ID +
'_wing_angles.pickle') == True:
wings = wing_detector.WingDetector(ZOOM_FMF, BAG_FILE, dtarget,
arenaCtr, wingImageDir)
wings.execute()
wings.wingData.to_pickle(MAIN_DIRECTORY + '/JAR/wing_data/' + FLY_ID +
'_wing_angles.pickle')
videoData = pd.read_pickle(MAIN_DIRECTORY + '/JAR/wing_data/' + FLY_ID +
'_wing_angles.pickle')
videoData[['Length', 'Width', 'Left',
'Right']] = videoData[['Length', 'Width', 'Left',
'Right']].astype(np.float64)
videoData = convert_timestamps(videoData)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
videoData['Laser0_state'] = laser_states['Laser0_state'].asof(
videoData.index).fillna(value=1)
videoData['Laser1_state'] = laser_states['Laser1_state'].asof(
videoData.index).fillna(value=0) #YAY!
videoData['Laser2_state'] = laser_states['Laser2_state'].asof(
videoData.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
videoData['Laser0_state'] = 0
videoData['Laser2_state'] = 0
videoData['Laser1_state'] = 0
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
videoData['fly_x'] = positions['fly_x'].asof(
videoData.index).fillna(value=0)
videoData['fly_y'] = positions['fly_y'].asof(
videoData.index).fillna(value=0)
videoData['dtarget'] = dTarget.asof(videoData.index).fillna(value=0)
videoData[
'Timestamp'] = videoData.index #silly pandas bug for subtracting from datetimeindex...
try:
videoData['synced_time'] = videoData['Timestamp'] - videoData.Timestamp[
(videoData.Laser2_state + videoData.Laser1_state) > 0.001].index[0]
except:
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame0. "
videoData['synced_time'] = videoData[
'Timestamp'] - videoData.Timestamp.index[0]
videoData.index = videoData.synced_time
videoData.index = pd.to_datetime(videoData.index)
### WING EXTENSION ###
videoData['maxWingAngle'] = get_absmax(videoData[['Left', 'Right']])
#videoData[videoData['maxWingAngle'] > 3.1] = np.nan
program = 'dark' #FIXME
if program == 'IRR':
BEGINNING = videoData.Timestamp[
videoData.synced_time >= -30000000000].index[
0] #videoData.Timestamp.index[0]
#FIRST_IR_ON = videoData.Timestamp[((videoData.Laser1_state > 0.001) & (videoData.synced_time >= -1))].index[0]
FIRST_IR_ON = videoData.Timestamp[videoData.synced_time >= 0].index[0]
#FIRST_IR_OFF = videoData.Timestamp[((videoData.Laser1_state > 0.001) & (videoData.synced_time <= 120))].index[-1]
FIRST_IR_OFF = videoData.Timestamp[
videoData.synced_time >= 60000000000].index[0]
RED_ON = videoData.Timestamp[videoData.Laser2_state > 0.001].index[0]
RED_OFF = videoData.Timestamp[videoData.Laser2_state > 0.001].index[-1]
SECOND_IR_ON = videoData.Timestamp[
videoData.synced_time >= 320000000000].index[0]
#SECOND_IR_ON = videoData.Timestamp[((videoData.Laser1_state > 0.001) & (videoData.synced_time >= 120))].index[0]
SECOND_IR_OFF = videoData.Timestamp[
videoData.Laser1_state > 0.001].index[-1]
END = videoData.Timestamp.index[-1]
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
BEGINNING,
FIRST_IR_ON,
'1-prestim',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
FIRST_IR_ON,
FIRST_IR_OFF,
'2-IR1',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
FIRST_IR_OFF,
RED_ON,
'3-post-IR1',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
RED_ON,
RED_OFF,
'4-red',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
RED_OFF,
SECOND_IR_ON,
'5-post-red',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
SECOND_IR_ON,
SECOND_IR_OFF,
'6-IR2',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
SECOND_IR_OFF,
END,
'7-post-IR2',
background=False)
if program == 'dark':
BEGINNING = videoData.Timestamp[
videoData.synced_time >= -30000000000].index[0]
print set(videoData.Laser0_state), set(videoData.Laser1_state), set(
videoData.Laser2_state)
STIM_ON = videoData.Timestamp[videoData.Laser1_state > 0.001].index[0]
STIM_OFF = videoData.Timestamp[
videoData.Laser1_state > 0.001].index[-1]
LIGHTS_OUT = videoData.Timestamp[
videoData.Laser0_state < 0.001].index[0]
LIGHTS_ON = videoData.Timestamp[
videoData.Laser0_state < 0.001].index[-1]
END = videoData.Timestamp.index[-1]
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
BEGINNING,
STIM_ON,
'1-prestim',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
STIM_ON,
STIM_OFF,
'2-stim',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
STIM_OFF,
LIGHTS_OUT,
'3-post-stim',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
LIGHTS_OUT,
LIGHTS_ON,
'4-DARK',
background=False)
targets.plot_trajectory_and_wingext(videoData,
BAG_FILE,
LIGHTS_ON,
END,
'7-light',
background=False)
targets.plot_trajectory_and_wingext(videoData, BAG_FILE)
### ABDOMINAL BENDING ###
videoData[videoData['Length'] >
110] = np.nan #discard frames with bogus length. *************
videoData[
videoData['Length'] < 60] = np.nan #discard frames with bogus length.
trace = plot_single_trace(videoData)
trace.savefig(MAIN_DIRECTORY + '/TRACES/' + FLY_ID + '.png')
plt.close('all')
###FIRST COURTSHIP BOUT AFTER STIMULUS###
#courted, latency = latency_measures(videoData)
if 'binsize' in globals():
videoData = bin_data(videoData, binsize)
videoData.to_pickle(MAIN_DIRECTORY + '/JAR/' + FLY_ID + '_' + binsize +
'_fly.pickle')
else:
return videoData, courted, latency
def sync_jaaba_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
JAABA_CSV = FMF_DIR + '/registered_trx.csv'
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, JAABA)
jaaba_data = pd.read_csv(
JAABA_CSV,
sep=',',
names=['Timestamp', 'Length', 'Width', 'Theta', 'Left', 'Right'],
index_col=False)
jaaba_data[['Length', 'Width', 'Left',
'Right']] = jaaba_data[['Length', 'Width', 'Left',
'Right']].astype(np.float64)
jaaba_data = convert_timestamps(jaaba_data)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
jaaba_data['Laser0_state'] = laser_states['Laser0_state'].asof(
jaaba_data.index).fillna(value=1)
jaaba_data['Laser1_state'] = laser_states['Laser1_state'].asof(
jaaba_data.index).fillna(value=0) #YAY!
jaaba_data['Laser2_state'] = laser_states['Laser2_state'].asof(
jaaba_data.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
jaaba_data['Laser0_state'] = 0
jaaba_data['Laser2_state'] = 0
jaaba_data['Laser1_state'] = 0
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
jaaba_data['fly_x'] = positions['fly_x'].asof(
jaaba_data.index).fillna(value=0)
jaaba_data['fly_y'] = positions['fly_y'].asof(
jaaba_data.index).fillna(value=0)
jaaba_data['dtarget'] = targets.get_dist_to_nearest_target(
BAG_FILE)['dtarget'].asof(jaaba_data.index).fillna(value=0)
jaaba_data[
'Timestamp'] = jaaba_data.index #silly pandas bug for subtracting from datetimeindex...
number_of_bouts, bout_duration, first_TS, last_TS = utilities.detect_stim_bouts(
jaaba_data, 'Laser1_state')
try:
jaaba_data['synced_time'] = jaaba_data['Timestamp'] - last_TS
except:
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame 1500. "
jaaba_data['synced_time'] = jaaba_data[
'Timestamp'] - jaaba_data.Timestamp.index[1500]
jaaba_data.index = jaaba_data.synced_time
jaaba_data.index = pd.to_datetime(jaaba_data.index)
### WING EXTENSION ###
jaaba_data['maxWingAngle'] = get_absmax(jaaba_data[['Left', 'Right']])
#jaaba_data[jaaba_data['maxWingAngle'] > 3.1] = np.nan
number_of_bouts, stim_duration, first_TS, last_TS = utilities.detect_stim_bouts(
jaaba_data, 'Laser2_state') #HACK DANNO
jaaba_data['stim_duration'] = stim_duration
BEGINNING = jaaba_data.Timestamp.index[0]
END = jaaba_data.Timestamp.index[-1]
print BEGINNING
print first_TS
print last_TS
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE, background=False)
targets.plot_trajectory_and_wingext(jaaba_data,
BAG_FILE,
BEGINNING,
first_TS,
'1-prestim',
background=False)
targets.plot_trajectory_and_wingext(jaaba_data,
BAG_FILE,
first_TS,
last_TS,
'2-red',
background=False)
targets.plot_trajectory_and_wingext(jaaba_data,
BAG_FILE,
last_TS,
END,
'3-post-red',
background=False)
targets.plot_trajectory_and_wingext(jaaba_data, BAG_FILE)
### ABDOMINAL BENDING ###
jaaba_data[jaaba_data['Length'] >
110] = np.nan #discard frames with bogus length. *************
jaaba_data[
jaaba_data['Length'] < 60] = np.nan #discard frames with bogus length.
trace = plot_single_trace(jaaba_data)
trace.savefig(JAABA + 'TRACES/' + FLY_ID + '.png')
plt.close('all')
if 'binsize' in globals():
jaaba_data = bin_data(jaaba_data, binsize)
jaaba_data.to_pickle(JAABA + 'JAR/' + FLY_ID + '_' + binsize +
'_fly.pickle')
else:
return jaaba_data, courted, latency
import flymad_jaaba.utilities as utilities import flymad_jaaba.target_detector as target_detector import matplotlib.pyplot as plt fmf_file = '/media/DBATH_5/150107/DB201-GP-costim_wide_20150107_174101.fmf' bagdir = '/media/DBATH_5/150107/BAGS' tempdir = '/groups/dickson/home/bathd/Desktop/150119' bag_file = utilities.match_fmf_and_bag(fmf_file, bagdir) targs = target_detector.TargetDetector(fmf_file, tempdir) targs.plot_trajectory_on_background(bag_file) positions = utilities.get_positions_from_bag(bag_file) distances = targs.get_dist_to_nearest_target(bag_file) plt.plot(distances) plt.show() plt.plot(positions.fly_x, positions.fly_y)#, alpha=distances.values) plt.show()
def sync_jaaba_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
DATADIR_CSV = FMF_DIR + '/registered_trx.csv'
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, DATADIR)
for x in glob.glob(FMF_DIR + '/*zoom*' + '*.fmf'):
ZOOM_FMF = x
if not os.path.exists(FMF_DIR + '/TRACKING_FRAMES') == True:
os.makedirs(FMF_DIR + '/TRACKING_FRAMES')
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
dtarget = targets.get_dist_to_nearest_target(BAG_FILE)['dtarget']
(arena_centre), arena_radius = targets._arena.circ
if MAKE_MOVIES:
TRACKING_DIRECTORY = FMF_DIR + '/TRACKING_FRAMES/'
else:
TRACKING_DIRECTORY = None
if os.path.exists(FMF_DIR + '/wingdata.pickle'):
datadf = pd.read_pickle(FMF_DIR + '/wingdata.pickle')
datadf.columns = [
'BodyAxis', 'leftAngle', 'leftWingLength', 'Length', 'rightAngle',
'rightWingLength', 'Timestamp', 'Width'
]
else:
try:
wings = wing_detector.WingDetector(ZOOM_FMF, BAG_FILE, dtarget,
arena_centre,
TRACKING_DIRECTORY)
wings.execute()
wings.wingData.columns = [
'BodyAxis', 'leftAngle', 'leftWingLength', 'Length',
'rightAngle', 'rightWingLength', 'Timestamp', 'Width'
]
wings.wingData.to_pickle(FMF_DIR + '/wingdata.pickle')
datadf = DataFrame(wings.wingData)
except:
print 'ERROR PROCESSING WING TRACKING...', FMF_DIR
return
datadf['Frame_number'] = datadf.index
datadf = convert_timestamps(datadf)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
datadf['Laser0_state'] = laser_states['Laser0_state'].asof(
datadf.index).fillna(value=1)
datadf['Laser1_state'] = laser_states['Laser1_state'].asof(
datadf.index).fillna(value=0) #YAY!
datadf['Laser2_state'] = laser_states['Laser2_state'].asof(
datadf.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
datadf['Laser0_state'] = 0
datadf['Laser2_state'] = 0
datadf['Laser1_state'] = 0
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
datadf['fly_x'] = positions['fly_x'].asof(datadf.index).fillna(value=0)
datadf['fly_y'] = positions['fly_y'].asof(datadf.index).fillna(value=0)
datadf['dcentre'] = np.sqrt((datadf['fly_x'] - arena_centre[0])**2 +
(datadf['fly_y'] - arena_centre[1])**2)
datadf['dtarget'] = targets.get_dist_to_nearest_target(
BAG_FILE)['dtarget'].asof(datadf.index).fillna(value=0)
datadf[
'Timestamp'] = datadf.index #silly pandas bug for subtracting from datetimeindex...
try:
datadf['synced_time'] = datadf['Timestamp'] - datadf.Timestamp[
(datadf.Laser2_state + datadf.Laser1_state) > 0.001].index[0]
except:
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame0. "
datadf['synced_time'] = datadf['Timestamp'] - datadf.Timestamp.index[0]
datadf.index = datadf.synced_time
datadf.index = pd.to_datetime(datadf.index)
### WING EXTENSION ###
datadf['maxWingAngle'] = get_absmax(datadf[['leftAngle', 'rightAngle']])
datadf['maxWingLength'] = get_absmax(
datadf[['leftWingLength', 'rightWingLength']])
#datadf[datadf['maxWingAngle'] > 3.1] = np.nan
"""
program = 'dark'
plt.plot(datadf.Timestamp, datadf.Laser0_state, 'b')
plt.plot(datadf.Timestamp, datadf.Laser1_state, 'k')
plt.plot(datadf.Timestamp, datadf.Laser2_state, 'r')
plt.show()
if program == 'IRR':
BEGINNING =datadf.Timestamp[datadf.synced_time >= -30000000000].index[0]#datadf.Timestamp.index[0]
#FIRST_IR_ON = datadf.Timestamp[((datadf.Laser1_state > 0.001) & (datadf.synced_time >= -1))].index[0]
FIRST_IR_ON = datadf.Timestamp[datadf.synced_time >= 0].index[0]
#FIRST_IR_OFF = datadf.Timestamp[((datadf.Laser1_state > 0.001) & (datadf.synced_time <= 120))].index[-1]
FIRST_IR_OFF = datadf.Timestamp[datadf.synced_time >= 60000000000].index[0]
RED_ON = datadf.Timestamp[datadf.Laser2_state > 0.001].index[0]
RED_OFF = datadf.Timestamp[datadf.Laser2_state > 0.001].index[-1]
SECOND_IR_ON = datadf.Timestamp[datadf.synced_time >=320000000000].index[0]
#SECOND_IR_ON = datadf.Timestamp[((datadf.Laser1_state > 0.001) & (datadf.synced_time >= 120))].index[0]
SECOND_IR_OFF = datadf.Timestamp[datadf.Laser1_state > 0.001].index[-1]
END = datadf.Timestamp.index[-1]
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, BEGINNING, FIRST_IR_ON, '1-prestim', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, FIRST_IR_ON, FIRST_IR_OFF, '2-IR1', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, FIRST_IR_OFF, RED_ON, '3-post-IR1', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, RED_ON,RED_OFF, '4-red', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE,RED_OFF, SECOND_IR_ON,'5-post-red', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE,SECOND_IR_ON,SECOND_IR_OFF,'6-IR2', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE,SECOND_IR_OFF,END,'7-post-IR2', background=False)
if program == 'dark':
BEGINNING =datadf.Timestamp[datadf.synced_time >= -30000000000].index[0]
print set(datadf.Laser0_state), set(datadf.Laser1_state), set(datadf.Laser2_state)
STIM_ON = datadf.Timestamp[datadf.Laser1_state > 0.001].index[0]
STIM_OFF = datadf.Timestamp[datadf.Laser1_state > 0.001].index[-1]
LIGHTS_OUT = datadf.Timestamp[datadf.Laser0_state < 0.001].index[0]
LIGHTS_ON = datadf.Timestamp[datadf.Laser0_state < 0.001].index[-1]
END = datadf.Timestamp.index[-1]
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, BEGINNING, STIM_ON, '1-prestim', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, STIM_ON,STIM_OFF, '2-stim', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, STIM_OFF, LIGHTS_OUT,'3-post-stim', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, LIGHTS_OUT, LIGHTS_ON,'4-DARK', background=False)
targets.plot_trajectory_and_wingext(datadf, BAG_FILE, LIGHTS_ON, END,'7-light', background=False)
"""
targets.plot_trajectory_and_wingext(datadf, BAG_FILE)
### ABDOMINAL BENDING ###
#datadf[datadf['Length'] > 110] = np.nan #discard frames with bogus length. *************
#datadf[datadf['Length'] < 60] = np.nan #discard frames with bogus length.
trace = plot_single_trace(datadf)
trace.savefig(DATADIR + 'TRACES/' + FLY_ID + '.png')
plt.close('all')
###FIRST COURTSHIP BOUT AFTER STIMULUS###
#courted, latency = latency_measures(datadf)
datadf.to_pickle(FMF_DIR + '/frame_by_frame_synced.pickle')
datadf.to_csv(FMF_DIR + '/frame_by_frame_synced.csv', sep=',')
if 'binsize' in globals():
datadf = bin_data(datadf, binsize)
datadf.to_pickle(DATADIR + 'JAR/' + FLY_ID + '_' + binsize +
'_fly.pickle')
else:
return datadf, courted, latency
def sync_jaaba_with_ros(FMF_DIR):
print "Processing: ", FMF_DIR
DATADIR_CSV = FMF_DIR + '/registered_trx.csv'
FLY_ID, FMF_TIME, GROUP = parse_fmftime(FMF_DIR)
BAG_FILE = match_fmf_and_bag(FMF_TIME)
WIDE_FMF = utilities.match_wide_to_zoom(FMF_DIR, DATADIR)
FLY_IDx = FMF_DIR.split('/')[-1]
EXP_ID, DATE, TIME = FLY_IDx.split('_', 4)[0:3]
for x in glob.glob(FMF_DIR + '/*zoom*' + '*.fmf'):
ZOOM_FMF = x
if not os.path.exists(FMF_DIR + '/TRACKING_FRAMES') == True:
os.makedirs(FMF_DIR + '/TRACKING_FRAMES')
# COMPUTE AND ALIGN DISTANCE TO NEAREST TARGET
if (os.path.exists(FMF_DIR +
'/frame_by_frame_synced.pickle')) and not RETRACK:
datadf = pd.read_pickle(FMF_DIR + '/frame_by_frame_synced.pickle')
exp_group = EXP_ID #QUICKFIX. FIXME
else:
targets = target_detector.TargetDetector(WIDE_FMF, FMF_DIR)
targets.plot_targets_on_background()
targets.plot_trajectory_on_background(BAG_FILE)
dtarget = targets.get_dist_to_nearest_target(BAG_FILE)['dtarget']
(arena_centre), arena_radius = targets._arena.circ
if MAKE_MOVIES:
TRACKING_DIRECTORY = FMF_DIR + '/TRACKING_FRAMES/'
else:
TRACKING_DIRECTORY = None
if (os.path.exists(FMF_DIR + '/wingdata.pickle')) and not RETRACK:
datadf = pd.read_pickle(FMF_DIR + '/wingdata.pickle')
datadf.columns = [
'BodyAxis', 'leftAngle', 'leftWingLength', 'Length',
'rightAngle', 'rightWingLength', 'target_angle_TTM',
'target_distance_TTM', 'Timestamp', 'Width'
]
else:
try:
wings = wing_detector.WingDetector(ZOOM_FMF, BAG_FILE, dtarget,
arena_centre, RETRACK,
TRACKING_DIRECTORY)
wings.execute()
wings.wingData.columns = [
'BodyAxis', 'leftAngle', 'leftWingLength', 'Length',
'rightAngle', 'rightWingLength', 'target_angle_TTM',
'target_distance_TTM', 'Timestamp', 'Width'
]
wings.wingData.to_pickle(FMF_DIR + '/wingdata.pickle')
wings.tracking_info.to_pickle(FMF_DIR +
'/tracking_info.pickle')
datadf = DataFrame(wings.wingData)
if MAKE_MOVIES:
utilities.call_command(
"ffmpeg -f image2 -r 15 -i " + TRACKING_DIRECTORY +
"_tmp%05d.png -vf scale=iw/2:-1 -vcodec mpeg4 -b 8000k -y "
+ FMF_DIR + "/tracked_movie.mp4;")
utilities.call_command("rm -r " + TRACKING_DIRECTORY)
#wings.make_movie(wings._tempdir, wings.fmf_file.rsplit('/',1)[0]+'/tracked_movie.mp4',15)
except Exception, e:
traceback.print_exc()
print 'ERROR PROCESSING WING TRACKING...', FMF_DIR
print str(e)
return
datadf['Frame_number'] = datadf.index
datadf = convert_timestamps(datadf)
# ALIGN LASER STATE DATA
laser_states = utilities.get_laser_states(BAG_FILE)
try:
datadf['Laser0_state'] = laser_states['Laser0_state'].asof(
datadf.index).fillna(value=1)
datadf['Laser1_state'] = laser_states['Laser1_state'].asof(
datadf.index).fillna(value=0) #YAY!
datadf['Laser2_state'] = laser_states['Laser2_state'].asof(
datadf.index).fillna(value=0)
except:
print "\t ERROR: problem interpreting laser current values."
datadf['Laser0_state'] = 0
datadf['Laser2_state'] = 0
datadf['Laser1_state'] = 0
positions = utilities.get_positions_from_bag(BAG_FILE)
positions = utilities.convert_timestamps(positions)
datadf['fly_x'] = positions['fly_x'].asof(
datadf.index).fillna(method='pad')
datadf['fly_y'] = positions['fly_y'].asof(
datadf.index).fillna(method='pad')
datadf['dcentre'] = np.sqrt((
(datadf['fly_x'] - arena_centre[0]) / 4.8)**2 + (
(datadf['fly_y'] - arena_centre[1]) / 5.2)**2)
datadf['dtarget_temp'] = targets.get_dist_to_nearest_target(
BAG_FILE)['dtarget'].asof(datadf.index).fillna(method='pad')
datadf.loc[datadf['dtarget_temp'] >= 6.0,
'target_distance_TTM'] = np.nan
datadf['dtarget'] = datadf['target_distance_TTM'].fillna(
datadf['dtarget_temp'])
datadf[
'Timestamp'] = datadf.index #silly pandas bug for subtracting from datetimeindex...
#number_of_bouts, bout_duration, first_TS, last_TS = utilities.detect_stim_bouts(datadf, 'Laser1_state')
number_of_bouts, stim_duration, first_TS, last_TS = utilities.detect_stim_bouts(
datadf, 'Laser2_state') #HACK DANNO
datadf['stim_duration'] = stim_duration
try:
if stim_duration >= 0:
datadf['synced_time'] = datadf['Timestamp'] - datadf.Timestamp[
(datadf.Laser2_state +
datadf.Laser1_state) > 0.001].index[0]
#elif stim_duration == 0:
# datadf['synced_time'] = datadf['Timestamp'] - datadf.Timestamp.index[0] - pd.to_datetime('60s')
except Exception, e:
traceback.print_exc()
print "WARNING: Cannot synchronize by stimulus. Setting T0 to frame 900. "
datadf['synced_time'] = datadf[
'Timestamp'] - datadf.Timestamp.index[900]
print str(e)