From 75988b01443c0dc23dff9d5672ab86d2a8dd954f Mon Sep 17 00:00:00 2001
From: Mehdi Seifi <mehdiseifi@gmail.com>
Date: Thu, 24 Oct 2024 12:40:00 +0200
Subject: [PATCH] updated notebook

---
 run_pipeline.ipynb                 | 19 +++++++++++++------
 src/featureforest/utils/extract.py |  2 +-
 2 files changed, 14 insertions(+), 7 deletions(-)

diff --git a/run_pipeline.ipynb b/run_pipeline.ipynb
index 868a5b3..4a47619 100755
--- a/run_pipeline.ipynb
+++ b/run_pipeline.ipynb
@@ -201,17 +201,17 @@
    "outputs": [],
    "source": [
     "# input image\n",
-    "data_path = \"../datasets/Johan/dino_host/test_substacks/Stack02_bin4_(1-3598-3).tif\"\n",
+    "data_path = \"../datasets/data.tif\"\n",
     "data_path = Path(data_path)\n",
     "print(f\"data_path exists: {data_path.exists()}\")\n",
     "\n",
     "# random forest model\n",
-    "rf_model_path = \"../datasets/Johan/dino_host/mito/host/rf_model.bin\"\n",
+    "rf_model_path = \"../datasets/rf_model.bin\"\n",
     "rf_model_path = Path(rf_model_path)\n",
     "print(f\"rf_model_path exists: {rf_model_path.exists()}\")\n",
     "\n",
     "# result folder\n",
-    "segmentation_dir = Path(\"../datasets/Johan/dino_host/mito/host/segmentation_result\")\n",
+    "segmentation_dir = Path(\"../datasets/segmentation_result\")\n",
     "segmentation_dir.mkdir(parents=True, exist_ok=True)\n",
     "\n",
     "# temporary storage path for saving extracted embeddings patches\n",
@@ -249,9 +249,16 @@
    "outputs": [],
    "source": [
     "with open(rf_model_path, mode=\"rb\") as f:\n",
-    "    rf_model = pickle.load(f)\n",
-    "    rf_model.set_params(verbose=0)\n",
-    "\n",
+    "    model_data = pickle.load(f)\n",
+    "# compatibility check for old format rf model\n",
+    "if isinstance(model_data, dict):  # noqa: SIM108\n",
+    "    # new format\n",
+    "    rf_model = model_data[\"rf_model\"]\n",
+    "else:\n",
+    "    # old format\n",
+    "    rf_model = model_data\n",
+    "\n",
+    "rf_model.set_params(verbose=0)\n",
     "rf_model"
    ]
   },
diff --git a/src/featureforest/utils/extract.py b/src/featureforest/utils/extract.py
index dbf37ee..3aa6e51 100644
--- a/src/featureforest/utils/extract.py
+++ b/src/featureforest/utils/extract.py
@@ -66,7 +66,7 @@ def get_slice_features(
 
     # get sam encoder output for image patches
     print("extracting features:")
-    for b_idx in np_progress(range(num_batches), desc="extracting feature:"):
+    for b_idx in np_progress(range(num_batches), desc="extracting feature"):
         print(f"batch #{b_idx + 1} of {num_batches}")
         start = b_idx * batch_size
         end = start + batch_size