resolve use_fold early in app

app.py

@@ -41,9 +41,30 @@ gamma = 0.4
 threshold = 0.09951 if (normalize_similarity == 'test-train') else 0.6 # 0.381
 knn = 300
 rst_names = 'convnext_base_384_in22ft1k_colab220 efnv1b7_colab216 hub_efnv2xl_v73'.split()
+use_fold = {
+    'efnv1b7_colab216': 4,
+    'efnv1b7_colab225': 1,
+    'efnv1b7_colab197': 0,
+    'efnv1b7_colab227': 5,
+    'efnv1b7_v72': 6,
+    'efnv1b7_colab229': 9,
+    'efnv1b6_colab217': 5,
+    'efnv1b6_colab218': 6,
+    'hub_efnv2xl_colab221': 8, 
+    'hub_efnv2xl_v69': 2,
+    'hub_efnv2xl_v73': 0,
+    'efnv1b6_colab226': 2,
+    'hub_efnv2l_v70': 3,
+    'hub_efnv2l_colab200': 2, 
+    'hub_efnv2l_colab199': 1,
+    'convnext_base_384_in22ft1k_v68': 0,
+    'convnext_base_384_in22ft1k_colab220': 9,
+    'convnext_base_384_in22ft1k_colab201': 3,  # new
+}
 cfg_files = [hf_hub_download(modelbox, f'{x}_config.json') for x in rst_names]
 emb_files = [hf_hub_download(modelbox, f'{x}_emb.npz') for x in rst_names]
 rst_files = [hf_hub_download(modelbox, f'{x}.h5') for x in rst_names]
+use_folds = [use_fold[x] for x in rst_names]
 n_models = len(rst_names)
 
 
@@ -77,7 +98,7 @@ def fast_yolo_crop(image):
 
 
 # Preload embeddings for known individuals
-comp_embeddings = get_comp_embeddings(emb_files)
+comp_embeddings = get_comp_embeddings(emb_files, use_folds)
 
 # Preload embedding models, input sizes
 K.clear_session()

utils.py

@@ -56,41 +56,20 @@ def get_train_idx(use_fold):
     
     return np.argsort(train_idx)
 
-use_fold = {
-    'efnv1b7_colab216_emb.npz': 4,
-    'efnv1b7_colab225_emb.npz': 1,
-    'efnv1b7_colab197_emb.npz': 0,
-    'efnv1b7_colab227_emb.npz': 5,
-    'efnv1b7_v72_emb.npz': 6,
-    'efnv1b7_colab229_emb.npz': 9,
-    'efnv1b6_colab217_emb.npz': 5,
-    'efnv1b6_colab218_emb.npz': 6,
-    'hub_efnv2xl_colab221_emb.npz': 8, 
-    'hub_efnv2xl_v69_emb.npz': 2,
-    'hub_efnv2xl_v73_emb.npz': 0,
-    'efnv1b6_colab226_emb.npz': 2,
-    'hub_efnv2l_v70_emb.npz': 3,
-    'hub_efnv2l_colab200_emb.npz': 2, 
-    'hub_efnv2l_colab199_emb.npz': 1,
-    'convnext_base_384_in22ft1k_v68_emb.npz': 0,
-    'convnext_base_384_in22ft1k_colab220_emb.npz': 9,
-    'convnext_base_384_in22ft1k_colab201_emb.npz': 3,  # new
-}
-
 
-def get_comp_embeddings(emb_files):
+def get_comp_embeddings(emb_files, use_folds):
     "Load embeddings for competition images [n_images, embedding_size]"
 
     comp_embeddings = []
 
-    for npz_file in emb_files:
+    for npz_file, use_fold in zip(emb_files, use_folds):
         # Get embeddings for all competition images
         d = np.load(str(npz_file))
         comp_train_emb = d['train']
         comp_test_emb = d['test']
         
         # Restore original order of comp_train_emb, targets (use targets as fingerprint-check)
-        comp_train_idx = get_train_idx(use_fold[npz_file])
+        comp_train_idx = get_train_idx(use_fold)
         comp_train_emb = comp_train_emb[comp_train_idx, :]
         comp_embs = np.concatenate([comp_train_emb, comp_test_emb], axis=0)
         assert comp_embs.shape == (n_images, embedding_size)