Merge pull request #62 from mGarbowski/plots

Plots

Author: mGarbowski

notebooks/11-rococo-train-test-split.ipynb

@@ -34,7 +34,7 @@
     "import os\n",
     "import shutil\n",
     "import random\n",
-    "import csv\n"
+    "import csv"
    ]
   },
   {
@@ -52,9 +52,8 @@
     }
    ],
    "source": [
-    "\n",
     "os.chdir(\"..\")\n",
-    "print(os.getcwd())\n"
+    "print(os.getcwd())"
    ]
   },
   {
@@ -98,7 +97,7 @@
     "frame_files = sorted(os.listdir(frames_dir))\n",
     "\n",
     "print(f\"Number of face files: {len(face_files)}\")\n",
-    "print(f\"Number of frame files: {len(frame_files)}\")    "
+    "print(f\"Number of frame files: {len(frame_files)}\")"
    ]
   },
   {
@@ -147,16 +146,16 @@
     "    train_frames = []\n",
     "    test_frames = []\n",
     "    train_face_set = set(face_id_from_filename(f) for f in train_faces)\n",
-    "    \n",
+    "\n",
     "    for frame in all_frames:\n",
     "\n",
     "        face_id = face_id_from_filename(frame)\n",
     "        if face_id in train_face_set:\n",
     "            train_frames.append(frame)\n",
     "        else:\n",
     "            test_frames.append(frame)\n",
-    "    \n",
-    "    return train_frames, test_frames\n"
+    "\n",
+    "    return train_frames, test_frames"
    ]
   },
   {
@@ -290,7 +289,7 @@
     "        src = os.path.join(frames_dir, f)\n",
     "        dst = os.path.join(train_frames_dir, f)\n",
     "        shutil.copyfile(src, dst)\n",
-    "    \n",
+    "\n",
     "    for f in te_frames:\n",
     "        src = os.path.join(frames_dir, f)\n",
     "        dst = os.path.join(test_frames_dir, f)\n",
@@ -328,7 +327,7 @@
     "for ratio in split_ratios:\n",
     "    split_dir = os.path.join(splits_root, f\"split_{int(ratio*100)}\")\n",
     "    os.makedirs(split_dir, exist_ok=True)\n",
-    "    create_partitioned_set(ratio, dataset_root, split_dir)\n"
+    "    create_partitioned_set(ratio, dataset_root, split_dir)"
    ]
   },
   {
@@ -353,14 +352,17 @@
     "    frame_files = sorted(os.listdir(frames_dir))\n",
     "    return face_files, frame_files\n",
     "\n",
+    "\n",
     "def get_matches(face_id, frame_files, n_pairs):\n",
     "    matches = [f for f in frame_files if face_id_from_filename(f) == face_id]\n",
     "    return random.sample(matches, n_pairs)\n",
     "\n",
+    "\n",
     "def get_mismatches(face_id, frame_files, n_pairs):\n",
     "    mismatches = [f for f in frame_files if face_id_from_filename(f) != face_id]\n",
     "    return random.sample(mismatches, n_pairs)\n",
     "\n",
+    "\n",
     "def create_pairs(face_files, frame_files, n_pairs_per_face):\n",
     "    match_pairs = []\n",
     "    mismatch_pairs = []\n",
@@ -403,16 +405,16 @@
     "print(f\"Number of faces: {len(faces)}\")\n",
     "print(f\"Number of frames: {len(frames)}\")\n",
     "\n",
-    "n_train = int(len(faces) * 2/3)\n",
+    "n_train = int(len(faces) * 2 / 3)\n",
     "n_val = len(faces) - n_train\n",
     "print(f\"Number of training faces: {n_train}\")\n",
     "print(f\"Number of validation faces: {n_val}\")\n",
     "\n",
     "train_faces = faces[:n_train]\n",
     "val_faces = faces[n_train:]\n",
     "\n",
-    "train_frames = frames[:n_train*31]\n",
-    "val_frames = frames[n_train*31:]\n",
+    "train_frames = frames[: n_train * 31]\n",
+    "val_frames = frames[n_train * 31 :]\n",
     "\n",
     "print(train_faces[-4:])\n",
     "print(val_faces[:4])\n",
@@ -1488,13 +1490,14 @@
    ],
    "source": [
     "def save_csv(pairs, filepath):\n",
-    "    with open(filepath, mode='w', newline='') as file:\n",
+    "    with open(filepath, mode=\"w\", newline=\"\") as file:\n",
     "        writer = csv.writer(file)\n",
     "        writer.writerow([\"face\", \"frame\"])\n",
     "        for face, frame in pairs:\n",
     "            writer.writerow([face, frame])\n",
     "    print(f\"Saved {len(pairs)} pairs to {filepath}\")\n",
     "\n",
+    "\n",
     "save_csv(match_train_pairs, \"data/rococo2v3-dev/train_match_pairs.csv\")\n",
     "save_csv(mismatch_train_pairs, \"data/rococo2v3-dev/train_mismatch_pairs.csv\")\n",
     "save_csv(match_val_pairs, \"data/rococo2v3-dev/val_match_pairs.csv\")\n",
@@ -1529,14 +1532,23 @@
     }
    ],
    "source": [
-    "used_frames = set([\n",
-    "    *(frame for _, frame in match_train_pairs),\n",
-    "    *(frame for _, frame in mismatch_train_pairs),\n",
-    "    *(frame for _, frame in match_val_pairs),\n",
-    "    *(frame for _, frame in mismatch_val_pairs),\n",
-    "])\n",
+    "used_frames = set(\n",
+    "    [\n",
+    "        *(frame for _, frame in match_train_pairs),\n",
+    "        *(frame for _, frame in mismatch_train_pairs),\n",
+    "        *(frame for _, frame in match_val_pairs),\n",
+    "        *(frame for _, frame in mismatch_val_pairs),\n",
+    "    ]\n",
+    ")\n",
     "\n",
-    "len(used_frames), sum((len(match_train_pairs), len(mismatch_train_pairs), len(match_val_pairs), len(mismatch_val_pairs)))"
+    "len(used_frames), sum(\n",
+    "    (\n",
+    "        len(match_train_pairs),\n",
+    "        len(mismatch_train_pairs),\n",
+    "        len(match_val_pairs),\n",
+    "        len(mismatch_val_pairs),\n",
+    "    )\n",
+    ")"
    ]
   },
   {