import os
import sys
import tensorflow as tf
from tensorflow.keras.callbacks import ModelCheckpoint
root_path = os.path.abspath(os.path.join('..'))
if root_path not in sys.path:
sys.path.append(root_path)
import config
from birds_loader import load_birds, build_dataset
from ssd_utils.ssd_loss import SSDLoss
from utils import import_by_name, train_test_split_tensors, MeanAveragePrecisionCallback
# Load train and validation data
train_image_paths, train_bnd_boxes = load_birds(split='train')
valid_image_paths, valid_bnd_boxes = load_birds(split='valid')
valid_data = build_dataset(valid_image_paths, valid_bnd_boxes,
image_size=config.IMAGE_SIZE,
batch_size=config.BATCH_SIZE)
for run in range(1, config.NUM_RUNS+1):
weights_dir = 'weights_{}'.format(run)
history_dir = 'history_{}'.format(run)
os.makedirs(weights_dir, exist_ok=True)
os.makedirs(history_dir, exist_ok=True)
for architecture in config.ARCHITECTURES:
model_class = import_by_name('ssd_utils.networks.' + architecture)
import pandas as pd
import os
import sys
import tensorflow as tf
root_path = os.path.abspath('..')
if root_path not in sys.path:
sys.path.append(root_path)
import config_naive_pasting_ablation as config
from birds_loader import load_birds, build_dataset
from ssd_utils import output_encoder
from ssd_utils.metrics import MeanAveragePrecision
from utils import import_by_name, MeanAveragePrecisionCallback
test_image_paths, test_bnd_boxes = load_birds(split='test')
test_data = build_dataset(test_image_paths,
test_bnd_boxes,
image_size=config.IMAGE_SIZE,
batch_size=config.BATCH_SIZE)
meanAP_metric = MeanAveragePrecision()
results = {'architecture': []}
for run in range(1, config.NUM_RUNS + 1):
results['run_{}'.format(run)] = []
for architecture in config.ARCHITECTURES:
model_class = import_by_name('ssd_utils.networks.' + architecture)
model = model_class(num_classes=len(config.CLASSES))
states = torch.load('cub_nets_state.pth')
opt = states['options']
MASK_SIZE = opt.sizex
if "netEncM" in states:
netEncM = models._netEncM(sizex=opt.sizex,
nIn=opt.nx,
nMasks=opt.nMasks,
nRes=opt.nResM,
nf=opt.nfM,
temperature=opt.temperature).to(device)
netEncM.load_state_dict(states["netEncM"])
netEncM.eval()
real_image_paths, bbox_crops = load_birds(root=REAL_OBJECTS_DIR, split='train')
def sample_object():
obj_idx = np.random.choice(len(real_image_paths))
image_path, bbox = real_image_paths[obj_idx], bbox_crops[obj_idx][0]
image = Image.open(image_path.numpy())
image = image.crop(bbox.numpy()[:-1])
image = image.resize((MASK_SIZE, MASK_SIZE))
image = np.array(image)
if image.ndim != 3: return sample_object()
return image
def calculate_mask(obj):
obj_t = obj.transpose(2, 0, 1)