From ff2dfbfee7d5fb103d9f7115341fc7ea72715d3a Mon Sep 17 00:00:00 2001 From: Nataniel Ruiz Date: Wed, 25 Dec 2019 20:23:14 -0400 Subject: [PATCH] next --- cyclegan/models/test_model.py | 26 -------------------------- 1 file changed, 26 deletions(-) diff --git a/cyclegan/models/test_model.py b/cyclegan/models/test_model.py index ea97866..6a4ff8a 100644 --- a/cyclegan/models/test_model.py +++ b/cyclegan/models/test_model.py @@ -91,32 +91,6 @@ class TestModel(BaseModel): d = (generated - generated_noattack).norm(float('-inf')) return l1, l2, l0, d - def attack(self, label, inst, image=None): - # Encode Inputs - image = Variable(image) if image is not None else None - input_label, inst_map, real_image, _ = self.encode_input(Variable(label), Variable(inst), image, infer=True) - - # Fake Generation - if self.use_features: - if self.opt.use_encoded_image: - # encode the real image to get feature map - feat_map = self.netE.forward(real_image, inst_map) - else: - # sample clusters from precomputed features - feat_map = self.sample_features(inst_map) - input_concat = torch.cat((input_label, feat_map), dim=1) - else: - input_concat = input_label - - # Attack - pgd_attack = attacks.LinfPGDAttack(model=self.netG) - black = np.zeros((1, 3, input_concat.size(2), input_concat.size(3))) - black = torch.FloatTensor(black).cuda() - # print(input_concat.size()) - input_adv, perturb = pgd_attack.perturb(input_concat, black) - - return input_adv, perturb - def optimize_parameters(self): """No optimization for test model.""" pass