def decode_larcv1_metamsg(metamsg, override_plane=None):
"""
decode meta messages, which look like: 'Plane 65535 (rows,cols) = (0,0) ... Left Top (0,0) ... Right Bottom (0,0)\n'
constructor:
ImageMeta(const double width=0., const double height=0.,
const size_t row_count=0., const size_t col_count=0,
const double origin_x=0., const double origin_y=0.,
const PlaneID_t plane=::larcv::kINVALID_PLANE)
"""
print "decoding meta message: ", metamsg.decode("ascii")
meta_nums = [int(x) for x in re.findall("\d+", metamsg.decode("ascii"))]
width = meta_nums[5] - meta_nums[3]
height = meta_nums[4] - meta_nums[6]
rows = meta_nums[1]
cols = meta_nums[2]
plane = meta_nums[0]
if override_plane is not None:
plane = override_plane
origin_x = meta_nums[3]
origin_y = meta_nums[4]
meta = larcv.ImageMeta(width, height, rows, cols, origin_x, origin_y,
plane)
return meta
def Get_crop_image(plane, x_2d, y_2d, ev_img):
meta_crop = larcv.ImageMeta(512, 512 * 6, 512, 512, 0, 8448, plane)
meta_origin_x, meta_origin_y = meta_origin_helper(x_2d, y_2d, verbose=1)
meta_crop.reset_origin(meta_origin_x, meta_origin_y)
img_vtx = ev_img.at(plane).crop(meta_crop)
img_vtx = larcv.as_ndarray(img_vtx)
img_vtx = np.where(img_vtx < 10, 0, img_vtx)
img_vtx = np.where(img_vtx > 500, 500, img_vtx)
return img_vtx
def decode_larcv2_evimage2d(io, producername, imgdata_np, imgmeta_np):
# make evcontainer
evout = io.get_data("image2d", producername)
# make meta
nimgs = imgdata_np.shape[1]
for i in xrange(nimgs):
nrows = Long(imgmeta_np[0, i, 0, 1])
ncols = Long(imgmeta_np[0, i, 0, 0])
planeid = Long(imgmeta_np[0, i, 0, 6])
lcvmeta = larcv.ImageMeta(imgmeta_np[0, i, 0, 2], imgmeta_np[0, i, 0,
3],
imgmeta_np[0, i, 0, 4], imgmeta_np[0, i, 0,
5], nrows,
ncols, planeid)
# convert image
outarr = np.flip(imgdata_np[0, i, :, :].transpose((1, 0)), 0)
lcvimg = larcv.as_image2d_meta(outarr, lcvmeta)
evout.append(lcvimg)
return
if not o.initialize():
sys.exit(ERROR_WRITE_INIT)
for idx in xrange(NUM_EVENT):
img = larcv.Image2D(10, 10)
for x in xrange(img.as_vector().size()):
img.set_pixel(x, x % 3)
event_image1 = o.get_data(larcv.kProductImage2D, "original")
event_image1.Append(img)
event_image2 = o.get_data(larcv.kProductImage2D, "target")
event_image2.Append(img)
roi = larcv.ROI()
#ImageMeta(const double width=0., const double height=0.,
# const size_t row_count=0., const size_t col_count=0,
# const double origin_x=0., const double origin_y=0.,
# const PlaneID_t plane=::larcv::kINVALID_PLANE)
meta = larcv.ImageMeta(3, 3, 3, 3, 2, -2, 0)
roi.AppendBB(meta)
event_roi = o.get_data(larcv.kProductROI, "roi")
event_roi.Append(roi)
o.set_id(0, 0, idx)
p.process(o)
o.save_entry()
idx += 1
p.finalize()
o.finalize()
def initialize(self):
self._last_entry = -1
self._event_keys = []
self._metas = []
# configure the input
from larcv import larcv
from ROOT import TChain
# set 2d vs. 3d functions
as_numpy_array = None
dtype_keyword = ''
if self._flags.DATA_DIM == 3:
as_numpy_voxels = larcv.fill_3d_voxels
as_numpy_pcloud = larcv.fill_3d_pcloud
dtype_keyword = 'sparse3d'
elif self._flags.DATA_DIM == 2:
as_numpy_voxels = larcv.fill_2d_voxels
as_numpy_pcloud = larcv.fill_2d_pcloud
dtype_keyword = 'sparse2d'
else:
print('larcv IO not implemented for data dimension',
self._flags.DATA_DIM)
raise NotImplementedError
ch_data = TChain('%s_%s_tree' % (dtype_keyword, self._flags.DATA_KEY))
ch_label = None
if self._flags.LABEL_KEY:
ch_label = TChain('%s_%s_tree' %
(dtype_keyword, self._flags.LABEL_KEY))
for f in self._flags.INPUT_FILE:
ch_data.AddFile(f)
if ch_label: ch_label.AddFile(f)
self._voxel = []
self._feature = []
self._label = []
br_data, br_label = (None, None)
event_fraction = 1. / ch_data.GetEntries() * 100.
total_point = 0.
for i in range(ch_data.GetEntries()):
if self._flags.LIMIT_NUM_SAMPLE > 0 and i > self._flags.LIMIT_NUM_SAMPLE:
break
ch_data.GetEntry(i)
if ch_label: ch_label.GetEntry(i)
if br_data is None:
br_data = getattr(
ch_data,
'%s_%s_branch' % (dtype_keyword, self._flags.DATA_KEY))
if ch_label:
br_label = getattr(
ch_label, '%s_%s_branch' %
(dtype_keyword, self._flags.LABEL_KEY))
# HACK that should go away when unifying 2d and 3d data reps...
data_tensor = br_data
label_tensor = br_label
if self._flags.DATA_DIM == 2:
data_tensor = br_data.as_vector().front()
label_tensor = br_label.as_vector().front()
num_point = data_tensor.as_vector().size()
if num_point < 10: continue
self._event_keys.append(
(br_data.run(), br_data.subrun(), br_data.event()))
# HACK that should go away when unifying 2d and 3d data reps...
if self._flags.DATA_DIM == 2:
self._metas.append(larcv.ImageMeta(label_tensor.meta()))
else:
self._metas.append(larcv.Voxel3DMeta(br_data.meta()))
np_voxel = np.zeros(shape=(num_point, self._flags.DATA_DIM),
dtype=np.int32)
as_numpy_voxels(data_tensor, np_voxel)
self._voxel.append(np_voxel)
np_feature = np.zeros(shape=(num_point, 1), dtype=np.float32)
as_numpy_pcloud(data_tensor, np_feature)
self._feature.append(np_feature)
if ch_label:
np_label = np.zeros(shape=(num_point, 1), dtype=np.float32)
as_numpy_pcloud(label_tensor, np_label)
np_label = np_label.reshape([num_point]) # - 1.
self._label.append(np_label)
total_point += num_point
sys.stdout.write(
'Processed %d%% ... %d MB\r' %
(int(event_fraction * i), int(total_point * 4 * 2 / 1.e6)))
sys.stdout.flush()
sys.stdout.write('\n')
sys.stdout.flush()
self._num_channels = self._voxel[-1].shape[-1]
self._num_entries = len(self._voxel)
# Output
if self._flags.OUTPUT_FILE:
import tempfile
cfg = '''
IOManager: {
Verbosity: 2
Name: "IOManager"
IOMode: 1
OutFileName: "%s"
InputFiles: []
InputDirs: []
StoreOnlyType: []
StoreOnlyName: []
}
'''
cfg = cfg % self._flags.OUTPUT_FILE
cfg_file = tempfile.NamedTemporaryFile('w')
cfg_file.write(cfg)
cfg_file.flush()
self._fout = larcv.IOManager(cfg_file.name)
self._fout.initialize()
# get predictions from gpu
ADCvalue_np = pred_ADCvalue.detach().cpu().numpy().astype(np.float32)
ADC_t = ADC_t.detach().cpu()
labelbasic_t = labelbasic_t.detach().cpu()
weights_t = weights_t.detach().cpu()
for ib in range(batch_size):
if ientry >= nevts:
# skip last portion of last batch
break
# get meta
inputmeta.read_entry(ientry)
ev_meta = inputmeta.get_data("image2d", "ADC")
outmeta = larcv.ImageMeta(496, 832, 496, 832, plane, 0, 0)
beforeloop = time.time()
#variables for accuracy check
within2 = 0.0
within5 = 0.0
within10 = 0.0
within20 = 0.0
chargetotal = 0.0
labelbasic_numpy = labelbasic_t.numpy()
weights_numpy = weights_t.numpy()
ADC_numpy = ADC_t.numpy()
#save a copy of labels for use in creating diff and threshold images
cfg3 = larcv.CreatePSetFromFile(sys.argv[4], "WireMask")
mask.configure(cfg3)
stream1.initialize()
stream2.initialize()
p.initialize()
mask.initialize()
if not o.initialize():
sys.exit(ERROR_WRITE_INIT)
for idx in xrange(NUM_EVENT):
#we have to make the image from ImageMeta if we are going to
#use channel status as it checks image2d.plane
meta1 = larcv.ImageMeta(10, 10, 10, 10, 0, 10, 0)
img1 = larcv.Image2D(meta1)
for x in xrange(img1.as_vector().size()):
if x % 2 == 0: img1.set_pixel(x, 10)
else: img1.set_pixel(x, 0)
meta2 = larcv.ImageMeta(10, 10, 10, 10, 0, 10, 0)
img2 = larcv.Image2D(meta2)
for x in xrange(img2.as_vector().size()):
img2.set_pixel(x, 10)
if (x / 10) % 2 == 0: img2.set_pixel(x, 0)
#Input stream 1
event_image1_tpc = o.get_data(
larcv.kProductImage2D, "stream1_tpc"
) # combined image is going to steal (std::move) this from me
track_img = larcv.Image2D(wholeview_v.at(1).meta())
bg_img = larcv.Image2D(wholeview_v.at(2).meta())
shower_img.paint(0)
track_img.paint(0)
bg_img.paint(0)
for out, img2d in zip(out_v, crop_img2d_v):
# threshold scores for better compression
showerslice = out[:, 1, :, :].reshape(512, 512)
trackslice = out[:, 2, :, :].reshape(512, 512)
bgslice = out[:, 0, :, :].reshape(512, 512)
meta = img2d.meta()
# back to image2d
shrmeta = larcv.ImageMeta(meta.width(), meta.height(), meta.rows(),
meta.cols(), meta.min_x(), meta.min_y(), 0)
trkmeta = larcv.ImageMeta(meta.width(), meta.height(), meta.rows(),
meta.cols(), meta.min_x(), meta.min_y(), 1)
bgmeta = larcv.ImageMeta(meta.width(), meta.height(), meta.rows(),
meta.cols(), meta.min_x(), meta.min_y(), 2)
showercrop = larcv.as_image2d_meta(showerslice, shrmeta)
shower_img.overlay(showercrop, larcv.Image2D.kOverWrite)
trackcrop = larcv.as_image2d_meta(trackslice, trkmeta)
track_img.overlay(trackcrop, larcv.Image2D.kOverWrite)
bgcrop = larcv.as_image2d_meta(bgslice, bgmeta)
bg_img.overlay(bgcrop, larcv.Image2D.kOverWrite)
h2d = larcv.as_th2d(shower_img, "shower_img")
c.Clear()
h2d.Draw("colz")
c.Update()
for ip,bbb in enumerate(bb_v):
print " plane ",ip,": ",bbb.dump()
isbad = True
bounds.append( (rmin,cmin,rmax,cmax) )
# we have to get the row, col bounds in the source image
if isbad:
#sign of bad image
image_meta.append(None)
target_meta.append(None)
continue
source_np[ib,0,:,:] = img_np[2,0,bounds[2][0]:bounds[2][2],bounds[2][1]:bounds[2][3]] # yplane
target_np[ib,0,:,:] = img_np[0,0,bounds[0][0]:bounds[0][2],bounds[0][1]:bounds[0][3]] # uplane
# store region of image
image_meta.append( larcv.ImageMeta( bb_v[2], 512, 832 ) )
target_meta.append( larcv.ImageMeta( bb_v[0], 512, 832 ) )
if verbose:
print "batch using ",len(image_meta)," slots"
# filled batch, make tensors
source_t = torch.from_numpy( source_np ).to(device=torch.device("cuda:1"))
target_t = torch.from_numpy( target_np ).to(device=torch.device("cuda:1"))
tformat = time.time()-tformat
timing["+++format"] += tformat
if verbose:
print "time to slice and fill tensor batch: ",tformat," secs"
# run model
trun = time.time()
subrun_vertex = ev_img.subrun()
event_vertex = ev_img.event()
index_array = di.query('run == {:2d} & subrun == {:2d} & event == {:2d}'.format(run_vertex,subrun_vertex,event_vertex)).index.values
x_2d = ROOT.Double()
y_2d = ROOT.Double()
vertex_index = index_array[0]
vx , vy ,vz = df['v_x'][vertex_index], df['v_y'][vertex_index], df['v_z'][vertex_index]
whole_img = ev_img.at(plane)
whole_image=larcv.as_ndarray(whole_img)
whole_img.reset_origin(0, 8448)
larcv.Project3D(whole_img.meta(), vx, vy, vz, 0, plane, x_2d, y_2d)
meta_crop = larcv.ImageMeta(512,512*6,512,512,0,0,plane)
meta_origin_x, meta_origin_y = u.Meta_origin_helper(x_2d, y_2d, verbose=0)
meta_crop.reset_origin(meta_origin_x, meta_origin_y)
image_vtx = ev_img.at(plane).crop(meta_crop)
img_vtx = larcv.as_ndarray(image_vtx)
img_vtx = np.where(img_vtx<10 ,0 ,img_vtx)
img_vtx = np.where(img_vtx>500,500,img_vtx)
if(entry <= training_number):
image2d_array_output = io_output.get_data(larcv.kProductImage2D, 'image2d_binary')
image_as_2d = larcv.as_image2d(img_vtx)
image2d_array_output.Append(image_as_2d)
io_output.set_id(100,1,event_out)
io_output.save_entry()
event_out += 1
# get data
#source,target1,target2,flow1,flow2,visi1,visi2,fvisi1,fvisi2,Wminx,Uminx,Vminx,srcmeta = prep_data( iovalid, "valid", batchsize_valid,
# IMAGE_WIDTH, IMAGE_HEIGHT, ADC_THRESH, DEVICE )
#source,target1,target2,flow1,flow2,visi1,visi2,fvisi1,fvisi2,Wminx,Uminx,Vminx,srcmeta = prep_data2( io, ientry, batchsize_valid,
# IMAGE_WIDTH, IMAGE_HEIGHT, ADC_THRESH, DEVICE )
source, target1, target2, flow1, flow2, visi1, visi2, fvisi1, fvisi2, Wminx, Uminx, Vminx, srcmeta = prep_data3(
ioserver, batchsize_valid, IMAGE_WIDTH, IMAGE_HEIGHT, ADC_THRESH,
DEVICE)
# make imagemeta for source
srcmetas = []
targetumetas = []
targetvmetas = []
for b in xrange(batchsize_valid):
meta = larcv.ImageMeta(srcmeta[b, 2, 0, 0], srcmeta[b, 2, 0, 1],
srcmeta[b, 2, 0, 2], srcmeta[b, 2, 0, 3],
IMAGE_HEIGHT, IMAGE_WIDTH, 2)
srcmetas.append(meta)
meta = larcv.ImageMeta(srcmeta[b, 0, 0, 0], srcmeta[b, 0, 0, 1],
srcmeta[b, 0, 0, 2], srcmeta[b, 0, 0, 3],
IMAGE_HEIGHT, IMAGE_WIDTH, 2)
targetumetas.append(meta)
meta = larcv.ImageMeta(srcmeta[b, 1, 0, 0], srcmeta[b, 1, 0, 1],
srcmeta[b, 1, 0, 2], srcmeta[b, 1, 0, 3],
IMAGE_HEIGHT, IMAGE_WIDTH, 2)
targetvmetas.append(meta)
# check image2d input
img_adc = larcv.as_image2d_meta(
source[0, 0, :, :].detach().cpu().numpy().transpose((1, 0)),
srcmetas[0])
o.set_out_file(OUT_FNAME)
if not o.initialize():
sys.exit(ERROR_WRITE_INIT)
p = larcv.VtxInRegion()
p.set_verbosity(2)
cfg = larcv.CreatePSetFromFile(sys.argv[1], "VtxInRegion")
p.configure(cfg)
x = []
y = []
for idx1 in xrange(0, 1000, 10):
for idx2 in xrange(-100, 100, 2):
event_roi1 = o.get_data(larcv.kProductROI, "vtxinregion")
bb1 = larcv.ImageMeta(2, 2, 2, 2, 2, 2, 0) #doesn't matter
roi1 = larcv.ROI()
roi1.Type(2)
roi1.Position(0, idx2, idx1, 0)
roi1.AppendBB(bb1)
event_roi1.clear()
event_roi1.Append(roi1)
if (p.process(o) == True):
x.append(idx1)
y.append(idx2)
import numpy as np
x = np.array(x)
y = np.array(y)
import matplotlib.pyplot as plt
