'conv2d_3': ['conv2_1.weight', 'conv2_1.bias'],
'conv2d_4': ['conv2_2.weight', 'conv2_2.bias'],
'dense_1': ['fc_layer1.weight', 'fc_layer1.bias'],
'dense_2': ['fc_layer2.weight', 'fc_layer2.bias'],
}
def init_weight(m):
if type(m) == nn.Linear:
torch.nn.init.xavier_normal_(m.weight)
#torch.nn.init.xavier_uniform_(m.bias)
if type(m) == nn.Conv2d:
torch.nn.init.xavier_normal_(m.weight)
#torch.nn.init.kaiming_uniform_(m.bias)
if __name__ == "__main__":
args = util.get_args(presets.PRESET)
with torch.cuda.device(2):
net = model.CifarNet_Vanilla()
if args.finetune:
net = util.load_latest_model(args, net)
else:
#net.apply(init_weight)
keras_model = get_keras_model()
model_path = os.path.join(os.getcwd(), 'test', 'models', "cifar10_cnn.h5")
net = weight_transfer.initialize_with_keras_hdf5(keras_model, map_dict, net, model_path)
omth_util.save_model(args, args.curr_epoch, net.state_dict())
#net.to(args.device)
net.cuda()
#summary(net, input_size=(3, 32, 32), device=device)
#train_set = fetch_data(args, [("data_batch_1", "data_batch_2", "data_batch_3", "data_batch_4", "data_batch_5")])
import os, time, datetime, random, glob
import cv2, torch
import numpy as np
import lp_data as data
import lp_model as model
from lp_args import parse_arguments
import omni_torch.utils as util
import lp_preset as preset
from imgaug import augmenters
import matplotlib.pyplot as plt
args = parse_arguments()
opt = util.get_args(preset.PRESET)
args = util.cover_edict_with_argparse(opt, args)
torch.backends.cudnn.benchmark = True
def test(net, test_sample, gt_label):
test_sample = test_sample.cuda()
pred, attn = net(test_sample, test=True)
if args.global_embedding:
l_pred = pred.unsqueeze(0).repeat(pred.shape[0], 1, 1) # * lr_attn
r_pred = pred.unsqueeze(1).repeat(1, pred.shape[0], 1) # * lr_attn
dis = torch.sum((l_pred - r_pred), dim=-1)
else:
l_attn = attn.unsqueeze(0).repeat(attn.shape[0], 1, 1, 1)
r_attn = attn.unsqueeze(1).repeat(1, attn.shape[0], 1, 1)
lr_attn = l_attn * r_attn
l_pred = pred.unsqueeze(0).repeat(pred.shape[0], 1, 1, 1) # * lr_attn
import omni_torch.visualize.basic as vb
import dd_data as data
import dd_preset as preset
import dd_model as model
from dd_loss import MultiBoxLoss
from dd_utils import *
from dd_preprocess import *
from dd_augment import *
from dd_postprocess import combine_boxes
from dd_vis import visualize_bbox, print_box
PIC = os.path.expanduser("~/Pictures/")
TMPJPG = os.path.expanduser("~/Pictures/tmp.jpg")
opt = preset.parse_arguments()
args = util.get_args(preset.PRESET, opt=opt)
cfg = model.cfg
cfg['super_wide'] = args.cfg_super_wide
cfg['super_wide_coeff'] = args.cfg_super_wide_coeff
cfg['overlap_thresh'] = args.jaccard_distance_threshold
def fit(args, cfg, net, detector, dataset, optimizer, is_train):
def avg(list):
return sum(list) / len(list)
if is_train:
net.train()
else:
net.eval()
Loss_L, Loss_C = [], []