def _generateBatchFromIDs(self, imageIdxs, batchIdx):
imagesID = [self.dataID[idx] for idx in imageIdxs]
imageMeta = self.imagesMeta[imagesID[0]]
imageMeta = cp.copy(imageMeta)
[dataXI, dataXH,
dataXO], _ = image.getXData(imageMeta, self.images_path, self.cfg,
batchIdx)
dataXW = image.getDataPairWiseStream(imageMeta, self.cfg)
dataXW = np.expand_dims(dataXW, axis=0)
y, _, _ = image.getYData(imagesID, self.imagesMeta, self.GTMeta,
self.cfg)
if dataXH.shape[1] > 32:
dataXH = dataXH[:, :32, :]
dataXO = dataXO[:, :32, :]
dataXW = dataXW[:, :32, :, :, :]
y = y[:, :32, :]
X = [dataXI, dataXH, dataXO, dataXW]
X = [X[i + 1] for i in range(len(X) - 1) if self.inputs[i]]
if self.inputs[0] or self.inputs[1]:
X = [dataXI] + X
return X, y
def _generateBatchFromIDs(self, batchID):
batchID = [self.dataID[idx] for idx in batchID]
# print(batchID)
[dataXP, dataXB] = image.getX2Data(batchID, self.imagesMeta,
self.images_path, self.cfg)
dataXW = image.getDataPairWiseStream(batchID, self.imagesMeta,
self.cfg)
X = [dataXP, dataXB, dataXW]
X = [X[i] for i in range(len(X)) if self.inputs[i]]
y, _ = image.getYData(batchID, self.imagesMeta, self.GTMeta, self.cfg)
return X, y
def __init__(self,
imagesMeta,
GTMeta,
cfg,
data_type='train',
do_meta=False):
'Initialization'
self.do_meta = do_meta
self.data_type = data_type
if data_type == 'train':
g_cfg = cfg.train_cfg
elif data_type == 'val':
g_cfg = cfg.val_cfg
else:
g_cfg = cfg.test_cfg
self.gen_type = g_cfg.type
self.batch_size = g_cfg.batch_size
self.nb_batches = g_cfg.nb_batches
self.images_per_batch = g_cfg.images_per_batch
self.nb_samples_per_image = int(self.batch_size /
self.images_per_batch)
self.shuffle = g_cfg.shuffle
self.inputs = cfg.inputs
cfg.img_out_reduction = (1, 1)
self.data_path = cfg.data_path
self.images_path = self.data_path + 'images/'
self.images_path = self.images_path + 'test/' if self.data_type == 'test' else self.images_path + 'train/'
self.nb_classes = cfg.nb_classes
self.cfg = cfg
self.dataID = list(imagesMeta.keys())
self.dataID.sort()
if self.shuffle:
r.shuffle(self.dataID)
self.imagesMeta = imagesMeta
self.GTMeta = GTMeta
gt_label, _, _ = image.getYData(self.dataID, self.imagesMeta,
self.GTMeta, self.cfg)
self.nb_images = len(self.dataID)
self.nb_samples = gt_label.shape[1]
if self.nb_batches is None:
self.nb_batches = m.ceil(self.nb_samples / self.batch_size)
def _generateBatchFromIDs(self, batchID):
batchID = [self.dataID[idx] for idx in batchID]
# print(batchID)
imageMeta = self.imagesMeta[batchID[0]]
imageMeta = cp.copy(imageMeta)
[dataXP, dataXB], img = image.getX2Data(imageMeta, self.images_path,
self.cfg)
dataXW = image.getDataPairWiseStream(imageMeta, self.cfg)
X = [dataXP, dataXB, dataXW]
X = [X[i] for i in range(len(X)) if self.inputs[i]]
y, _, _ = image.getYData(batchID, self.imagesMeta, self.GTMeta,
self.cfg)
y = y[0]
if self.do_meta:
return X, y, imageMeta, img
return X, y
def getYData(self):
gt_label, _, _ = image.getYData(self.dataID, self.imagesMeta,
self.GTMeta, self.cfg, 32)
gt_label = gt_label[0, :, :]
return gt_label
sys.path.append('../../classification/models/')
sys.path.append('../../shared/')
import utils
from model_trainer import model_trainer
from load_data import data
from generators import DataGenerator
import numpy as np
import metrics
import image
if False:
# Load data
print('Loading data...')
data = data(method='normal')
cfg = data.cfg
genTest = DataGenerator(imagesMeta=data.testMeta,
GTMeta=data.testGTMeta,
cfg=cfg,
data_type='test')
gt_label, _, _ = image.getYData(genTest.dataID, genTest.imagesMeta,
genTest.GTMeta, genTest.cfg)
path = 'C:\\Users\\aag14/Documents/Skole/Speciale/results/HICO/hoi80/yhat1'
yhat = utils.load_obj(path)
evalHOI = metrics.EvalResults(None, genTest, yhat=yhat, y=gt_label[0])
print(evalHOI.mAP, evalHOI.F1)
trainer_th = model_trainer(model=model_th_all,
genTrain=genTrain,
genVal=None,
genTest=None,
task=cfg.task)
trainer_th.compileModel(cfg)
if False:
imageID = 'HICO_train2015_00000015.jpg'
[dataXP_new,
dataXB_new] = image.getX2Data([imageID], genTrain.imagesMeta,
genTrain.images_path, genTrain.cfg)
dataXW_new = image.getDataPairWiseStream([imageID], genTrain.imagesMeta,
genTrain.cfg)
y_new, _, _ = image.getYData([imageID], genTrain.imagesMeta,
genTrain.GTMeta, genTrain.cfg)
dataXP_new = dataXP_new[0]
dataXB_new = dataXB_new[0]
dataXW_new = dataXW_new[0]
y_new = y_new[0]
# trainer_th.trainModel(cfg)
# print('Testing model on test...')
# resTest = trainer_th.evaluateModel(genTest)
# print("F1 (test!):", resTest.F1, "nb_zeros", resTest.nb_zeros)
j = 0
for i in range(1):
X, y = next(trainIterator)
