def get_similar_image(gid):
global images
if len(gid_list) == 0:
return jsonify(finallyDone=True)
index = None
length = len(listCursor)
#Loop to find location of given gid in sorted list
for i in range(length):
if listCursor[i]['gid'] == gid:
index = i
break
closeImages = []
key = 1
while len(closeImages) < 10:
#Looping till an unlabeled image found
option1 = listCursor[min(index+key,length-1)]
option2 = listCursor[max(index-key,0)]
if option1['gid'] in gid_list and not images[option1['gid']][1] and min(index+key,length-1) not in closeImages:
closeImages.append(min(index+key,length-1))
if option2['gid'] in gid_list and not images[option2['gid']][1] and max(index-key,0) not in closeImages:
closeImages.append(max(index-key,0) )
key += 1
entropyDistance = [(listCursor[x]['entropy']-listCursor[index]['entropy'])**2 for x in closeImages]
closest = np.argmin(entropyDistance)
nextGid = listCursor[closeImages[closest]]['gid']
images[nextGid][1] = True
#Opening next file to read info from
fileName = 'annotation_info/' + ibs.get_image_gnames(nextGid) + '.JSON'
with open(fileName) as data_file:
jsonData = json.load(data_file)
src = return_src(ibs.get_image_paths(nextGid))
img = ibs.get_images(nextGid)
return jsonify(image=src,id=nextGid,dim1=img.shape[0],dim2=img.shape[1],FinallyDone=False,data=jsonData,totalImages=totalImages,imagesLeft=len(gid_list))
def runImage(gid,ibs,network_fn):
imageLocations = "/home/zach/data/Flukes/CRC_combined constrained/"
name = imageLocations+ibs.get_image_gnames(gid)
img = cv2.imread(name)
img_normed = prepare_img(img)
img_output = network_fn(img_normed)
img_output = img_output.swapaxes(1,3)[0]
args = np.argmax(img_output, axis=2)
background_mask = args == 2
whale_mask = args == 1
seam_mask = args == 0
img_output[:,:][background_mask] = [255,0,0]
img_output[:,:][whale_mask] = [0,255,255]
img_output[:,:][seam_mask] = [255,0,255]
gradientImg = np.copy(img_output)
gradientImg[:,:][background_mask] = 255
gradientImg[:,:][whale_mask] = 0
gradientImg[:,:][seam_mask] = 255
gradientImg = gradientImg.astype(np.uint8)
gradientvalue = float(np.average(np.absolute(cv2.Laplacian(gradientImg,cv2.CV_64F))))
entrpy = entropy(img_output.swapaxes(0,2))
entrpy = np.average(entrpy)
bins = []
bins.append(np.sum(background_mask))
bins.append(np.sum(whale_mask))
bins.append(np.sum(seam_mask))
cv2.imwrite('tmp.png',img_output)
src = return_src('tmp.png')
return {'bins':bins,'gid':gid,'png':src,'version':nextVersion,'entropy':float(entrpy),'gradient':gradientvalue}
def get_next_image():
global index
global images
if len(gid_list) == 0:
return jsonify(finallyDone=True)
while images[gid_list[index]][1]:
index += 1
if index == len(gid_list):
index = 0
gid = gid_list[index]
#Opening stored data from JSON file
fileName = 'annotation_info/' + ibs.get_image_gnames(gid) + '.JSON'
with open(fileName) as data_file:
jsonData = json.load(data_file)
src = return_src(ibs.get_image_paths(gid))
img = ibs.get_images(gid)
images[gid_list[index]][1] = True
index += 1
if index == len(gid_list):
index == 0
#sending information to client
return jsonify(image=src,id=gid,dim1=img.shape[0],dim2=img.shape[1],FinallyDone=False,data=jsonData,totalImages=totalImages,imagesLeft=len(gid_list))
