Skip to content
Snippets Groups Projects
Commit 7ee284b7 authored by Unknown's avatar Unknown
Browse files

added the ability to hot-swap localizer models

parent 3b6917c3
Branches
No related merge requests found
Hide whitespace changes
Inline Side-by-side
__pycache__/localizer.cpython-36.pyc
+ 0
− 0
View file @ 7ee284b7
No preview for this file type
gif_writer.ipynb
+ 7
− 9
View file @ 7ee284b7
%% Cell type:code id: tags:
``` python
import imageio
import os
import cv2
from datetime import datetime
from tqdm import tqdm_notebook as tqdm
```
%% Cell type:code id: tags:
``` python
in_dir = r'C:\Users\Wheeler\Desktop\LCL_software\Experiments\experiment_18_07_2018___12.20.17.722253'
out_loc = r'C:\Users\Wheeler\Desktop\LCL_software\Experiments\experiment_18_07_2018___12.20.17.avi'
in_dir = r'C:\Users\Wheeler\Desktop\LCL_software\Experiments\experiment_24_07_2018___14.41.25.882424'
out_loc = r'C:\Users\Wheeler\Desktop\LCL_software\Experiments\experiment_24_07_2018___14.41.25.avi'
images = []
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8,8))
files = [os.path.join(in_dir, f) for f in os.listdir(in_dir) if '.tif' in f]
def get_time(file_name):
# t = file_name.split('_')[-2].split('_')[0]
t=file_name.split('_')[-1].split('.tif')[0]
t = datetime.strptime(t, '%H.%M.%S.%f')
t = t.hour*3600 + t.minute*60 + t.second + t.microsecond/10**6
return t
files.sort(key=get_time,reverse = False)
# for file in files:
# print('processing...',file.split('/')[-1])
# img = cv2.imread(file,0)
# img = cv2.resize(img,(int(img.shape[1]/3),int(img.shape[0]/3)),interpolation = cv2.INTER_CUBIC)
# img = clahe.apply(img)
# images.append(img)
# images = [images[0]] * 2 + images
# imageio.mimsave(out_loc, images,duration=.2)
# print('done')
```
%% Cell type:code id: tags:
``` python
img = cv2.imread(files[0],1)
width = img.shape[1]
height = img.shape[0]
print(img.shape)
use_clahe = False
fourcc = cv2.VideoWriter_fourcc(*'MJPG') # Be sure to use lower case
out = cv2.VideoWriter(out_loc, fourcc, 14.0, (int(width/1), int(height/1)))
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8,8))
for file in tqdm(files):
img = cv2.imread(file,1)
img = cv2.resize(img,(int(width/1),int(height/1)))
if use_clahe == True:
r,g,b = img[:,:,0],img[:,:,1],img[:,:,2]
r = clahe.apply(r)
g = clahe.apply(g)
b = clahe.apply(b)
img[:,:,0] = r
img[:,:,1] = g
img[:,:,2] = b
out.write(img)
out.release()
print('done')
```
%% Output
(822, 1024, 3)
done
%% Cell type:code id: tags:
``` python
```
......
localizer.py
+ 29
− 31
View file @ 7ee284b7
......@@ -11,6 +11,7 @@ from PyQt5.QtWidgets import QApplication
import pickle
from sklearn.preprocessing import StandardScaler
from utils import MeanIoU
from keras import backend as K
graph = tf.get_default_graph()
num_classes = 3
......@@ -27,41 +28,46 @@ class Localizer(QtCore.QObject):
ai_fire_qswitch_signal = QtCore.pyqtSignal('PyQt_PyObject')
start_laser_flash_signal = QtCore.pyqtSignal()
qswitch_screenshot_signal = QtCore.pyqtSignal('PyQt_PyObject')
# qswitch_screenshot_signal = QtCore.pyqtSignal()
def __init__(self, parent = None):
super(Localizer, self).__init__(parent)
# self.localizer_model = load_model(os.path.join(experiment_folder_location,'multiclass_localizer18_2.hdf5'),custom_objects={'mean_iou': mean_iou})
self.localizer_model = load_model(os.path.join(experiment_folder_location,'binary_green_hope_localizer3.hdf5'))
# self.localizer_model = load_model(os.path.join(experiment_folder_location,'multiclass_localizer14.hdf5'))
# self.localizer_model = load_model(os.path.join(experiment_folder_location,'binary_localizer6.hdf5'))
self.norm = StandardScaler()
self.hallucination_img = cv2.imread(os.path.join(experiment_folder_location,'before_qswitch___06_07_2018___11.48.59.274395.tif'))
self.localizer_model = load_model(os.path.join(experiment_folder_location,'multiclass_localizer18_2.hdf5'),custom_objects={'mean_iou': mean_iou})
self.norm = StandardScaler()
self.localizer_model._make_predict_function()
self.position = np.zeros((1,2))
self.well_center = np.zeros((1,2))
self.lysed_cell_count = 0
self.get_well_center = True
self.delay_time = .2
self.delay_time = .35
self.cells_to_lyse = 1
self.cell_type_to_lyse = 'red'
self.lysis_mode = 'direct'
self.auto_lysis = False
img = self.get_network_output(self.hallucination_img,'binary')
print('IMG SHAPE:',img.shape)
cv2.imshow('img',self.get_network_output(self.hallucination_img,'binary'))
# self.hallucination_img = cv2.imread(os.path.join(experiment_folder_location,'before_qswitch___06_07_2018___11.48.59.274395.tif'))
# img = self.get_network_output(self.hallucination_img,'binary')
# print('IMG SHAPE:',img.shape)
# cv2.imshow('img',img)
def stop_auto_lysis(self):
self.auto_lysis = False
def change_type_to_lyse(self,index):
map_dict = {
0:'red',
1:'green',
2:'green hope'
0:('red','multiclass_localizer18_2.hdf5'),
1:('green','multiclass_localizer18_2.hdf5'),
2:('green hope','binary_green_hope_localizer3.hdf5')
}
self.cell_type_to_lyse = map_dict[index]
self.cell_type_to_lyse = map_dict[index][0]
comment('loading cell localizer model...{}'.format(map_dict[index][1]))
self.localizer_model = None
K.clear_session()
# TODO: remove this if statement once we have a model that doesnt need the custom object
if map_dict[index] != 'green hope':
self.localizer_model = load_model(os.path.join(experiment_folder_location,map_dict[index][1]),custom_objects={'mean_iou': mean_iou})
else:
self.localizer_model = load_model(os.path.join(experiment_folder_location,map_dict[index][1]))
self.localizer_model._make_predict_function()
comment('changed cell type to:'+str(self.cell_type_to_lyse))
def change_lysis_mode(self,index):
......@@ -148,13 +154,15 @@ class Localizer(QtCore.QObject):
self.stop_laser_flash_signal.emit()
return
if self.lysed_cell_count >= self.cells_to_lyse:
self.delay()
self.return_to_original_position(self.well_center)
return
time.sleep(.2)
self.delay()
self.move_frame(let)
time.sleep(.2)
self.delay()
QApplication.processEvents()
self.delay()
QApplication.processEvents()
time.sleep(.2)
self.lyse_all_in_view()
self.return_to_original_position(self.well_center)
......@@ -226,10 +234,6 @@ class Localizer(QtCore.QObject):
elif lyse_type == 'excision':
self.excision_lysis(cell_contours)
# def get_return_vec(self,contour):
# return_vec = np.zeros((1,2))
# for vec in contour
def excision_lysis(self,cell_contours):
# for each contour we want to trace it
window_center = np.array([125./2,125./2])
......@@ -245,7 +249,6 @@ class Localizer(QtCore.QObject):
else:
# return_vec = np.sum((cell_contours[i-1].reshape(-1,2)),axis = 1)
new_center = contour[0].reshape(2)
print('hitting!',return_vec.shape,new_center.shape)
self.move_to_target(-return_vec.reshape(2) - contour_start + new_center,False)
old_center = new_center
contour_start = np.copy(new_center)
......@@ -268,7 +271,7 @@ class Localizer(QtCore.QObject):
self.lysed_cell_count += 1
self.qswitch_screenshot_signal.emit(2)
self.ai_fire_qswitch_signal.emit(False)
time.sleep(.1)
self.delay()
if self.auto_lysis == False:
self.stop_laser_flash_signal.emit()
return
......@@ -335,22 +338,17 @@ class Localizer(QtCore.QObject):
_,confidence_image = cv2.threshold(segmented_image[:,:,2],.5,1,cv2.THRESH_BINARY)
elif cell_type == 'green hope':
# assumes a binary image!
# TODO: find the optimal location on the AUC curve for the threshold
_,confidence_image = cv2.threshold(segmented_image,.3,1,cv2.THRESH_BINARY)
return confidence_image
def move_to_target(self,center,goto_reticle = False):
# we need to scale our centers up to the proper resolution and then
# send it to the stage
# localizer_move_slot(self, move_vector, goto_reticle = False,move_relative = True,scale_vector = True):
x = center[0]*851/125
y = center[1]*681/125
self.localizer_move_signal.emit(np.array([x,y]),goto_reticle,True,True)
# self.delay()
# self.ai_fire_qswitch_signal.emit(num_frames)
# self.delay()
# for i in range(1):
# self.ai_fire_qswitch_signal.emit()
# time.sleep(.1)
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment