diff options
| author | Jules Laplace <julescarbon@gmail.com> | 2020-02-18 18:29:03 +0100 |
|---|---|---|
| committer | Jules Laplace <julescarbon@gmail.com> | 2020-02-18 18:29:03 +0100 |
| commit | 75fa7f62aa9dbbbe3d69d03ad243a63a4b17c192 (patch) | |
| tree | 6bc8c1e56a4d11387ea048bcda55405fa13843e6 | |
| parent | fba1ca15295914e2d04bd91a1bcf5f23a68725e0 (diff) | |
simple sum
| -rw-r--r-- | cli/app/search/search_dense.py | 11 |
1 files changed, 8 insertions, 3 deletions
diff --git a/cli/app/search/search_dense.py b/cli/app/search/search_dense.py index ed22557..f52954c 100644 --- a/cli/app/search/search_dense.py +++ b/cli/app/search/search_dense.py @@ -231,9 +231,11 @@ def find_dense_embedding_for_images(params, opt_tag="inverse_" + timestamp(), op # print("Unknown feature extractor") # return + ################################################ # Inception feature extractor - - feature_extractor = hub.Module(str(params.feature_extractor_path)) + ################################################ + # feature_extractor = hub.Module(str(params.feature_extractor_path)) + feature_extractor = hub.Module("https://tfhub.dev/google/imagenet/inception_v3/feature_vector/1") feature_loss = feature_loss_tfhub height, width = hub.get_expected_image_size(feature_extractor) @@ -245,6 +247,9 @@ def find_dense_embedding_for_images(params, opt_tag="inverse_" + timestamp(), op # feat_loss_d, feat_err_d = feature_loss(feature_extractor, opt_feature_layers, BATCH_SIZE, gen_img_ch, target_img_ch, img_w - width, img_w - width, height, width) # feat_loss_e, feat_err_e = feature_loss(feature_extractor, opt_feature_layers, BATCH_SIZE, gen_img_ch, target_img_ch, int((img_w - width) / 2), int((img_w - width) / 2), height, width) + ################################################ + # VGG feature extractor + ################################################ model_path = os.path.join(app_cfg.DIR_NETS, 'vgg_16.ckpt') # conv1_1, conv1_2, conv3_2, conv4_2 opt_feature_layers = [ @@ -295,7 +300,7 @@ def find_dense_embedding_for_images(params, opt_tag="inverse_" + timestamp(), op # feat_loss_quint = tf.constant(0.0) # img_feat_err_quint = tf.constant(0.0) - img_rec_err = params.lambda_mse * img_mse_err + params.lambda_feat * img_feat_err + # img_rec_err = params.lambda_mse * img_mse_err + params.lambda_feat * img_feat_err inv_loss = (params.lambda_mse * mse_loss + params.lambda_feat * feat_loss) inv_loss_quad = (params.lambda_mse * mse_loss_quad + params.lambda_feat * feat_loss_quad) # inv_loss_quint = params.lambda_mse * mse_loss_quint + params.lambda_feat * feat_loss_quint |