Merge branch 'main' into roicluster

Author: danellecline

examples/sahi_detection.ipynb

@@ -4,10 +4,7 @@
    "cell_type": "markdown",
    "id": "5fec46c020b895a",
    "metadata": {
-    "collapsed": false,
-    "jupyter": {
-     "outputs_hidden": false
-    }
+    "collapsed": false
    },
    "source": [
     "# Welcome to the SAHI detection notebook. \n",
@@ -19,10 +16,7 @@
    "cell_type": "markdown",
    "id": "3ce3d68ddd33bd3c",
    "metadata": {
-    "collapsed": false,
-    "jupyter": {
-     "outputs_hidden": false
-    }
+    "collapsed": false
    },
    "source": [
     "# First, import needed dependencies"
@@ -37,7 +31,6 @@
    "source": [
     "import sys\n",
     "sys.path.append('..')\n",
-    "from pathlib import Path\n",
     "from huggingface_hub import hf_hub_download\n",
     "from pathlib import Path\n",
     "from sdcat.detect.sahi_detector import run_sahi_detect\n",
@@ -48,10 +41,7 @@
    "cell_type": "markdown",
    "id": "a366e55bf59e7a79",
    "metadata": {
-    "collapsed": false,
-    "jupyter": {
-     "outputs_hidden": false
-    }
+    "collapsed": false
    },
    "source": [
     "# Run detection\n",
@@ -65,10 +55,7 @@
    "id": "9f9dd3f8-c886-4d32-a203-5204a343e721",
    "metadata": {
     "collapsed": false,
-    "is_executing": true,
-    "jupyter": {
-     "outputs_hidden": false
-    }
+    "is_executing": true
    },
    "outputs": [
     {
@@ -172,10 +159,7 @@
    "cell_type": "markdown",
    "id": "3bda2110765d92e2",
    "metadata": {
-    "collapsed": false,
-    "jupyter": {
-     "outputs_hidden": false
-    }
+    "collapsed": false
    },
    "source": [
     "# Results"
@@ -185,10 +169,7 @@
    "cell_type": "markdown",
    "id": "ae874cee7b4e83a9",
    "metadata": {
-    "collapsed": false,
-    "jupyter": {
-     "outputs_hidden": false
-    }
+    "collapsed": false
    },
    "source": [
     "Results are saved to the **csv_out_path**. Let's take a look at the first few rows of the dataframe."

sdcat/cluster/cluster.py

@@ -10,14 +10,19 @@
 import json
 import seaborn as sns
 import numpy as np
-from matplotlib import pyplot as plt
 from umap import UMAP
 from hdbscan import HDBSCAN
 from sklearn.metrics.pairwise import cosine_similarity
+from sklearn.preprocessing import MinMaxScaler
 from sdcat.logger import info, warn, debug, err
 from sdcat.cluster.utils import cluster_grid, crop_square_image, square_image
 from sdcat.cluster.embedding import fetch_embedding, has_cached_embedding, compute_norm_embedding
 
+if find_spec("multicore_tsne"):
+    from multicore_tsne import MulticoreTSNE as TSNE
+else:
+    from sklearn.manifold import TSNE
+
 if find_spec("cuml"):
     info('=======> USING GPU for HDBSCAN AND UMAP <=========')
     from cuml.cluster import HDBSCAN as cuHDBSCAN  # pylint: disable=E0611, E0401
@@ -75,10 +80,9 @@ def _run_hdbscan_assign(
     # Get the number of samples which is the number of rows in the dataframe - this is used mostly for calculating coverage
     num_samples = df.shape[0]
 
-    # Compute the cosine similarity matrix
-    cosine_sim_matrix = cosine_similarity(df.values)
-    distance_matrix = 1 - cosine_sim_matrix
-    x = distance_matrix.astype(np.float64)
+    tsne = TSNE(n_components=2, perplexity=40, metric="cosine", n_jobs=8, random_state=42, verbose=True)
+    embedding = tsne.fit_transform(df.values)
+    x = MinMaxScaler().fit_transform(embedding) # scale the embedding to 0-1
 
     # Cluster the embeddings using HDBSCAN
     if have_gpu:
@@ -93,7 +97,7 @@ def _run_hdbscan_assign(
         labels = scan.fit_predict(x)
     else:
         scan = HDBSCAN(
-            metric='precomputed',
+            metric='l2',
             allow_single_cluster=True,
             min_cluster_size=min_cluster_size,
             min_samples=min_samples,
@@ -206,6 +210,7 @@ def _run_hdbscan_assign(
     df = pd.DataFrame({'x': xx[clustered, 0], 'y': xx[clustered, 1], 'labels': labels[clustered]})
     p = sns.jointplot(data=df, x='x', y='y', hue='labels')
     p.savefig(f"{out_path}/{prefix}_summary.png")
+    info(f"Saved {out_path}/{prefix}_summary.png")
 
     with open(f'{out_path}/{prefix}_summary.json', 'w') as f:
         json.dump(params, f)

sdcat/cluster/commands.py

@@ -260,8 +260,7 @@ def is_day(utc_dt):
         shutil.copy(Path(config_ini), save_dir / f'{prefix}_config.ini')
     else:
         warn(f'No detections found to cluster')
-
-
+ 
 @click.command('roi', help='Cluster roi. See cluster --config-ini to override cluster defaults.')
 @common_args.config_ini
 @click.option('--roi-dir', help='Input folder(s) with raw ROI images', multiple=True)
@@ -272,8 +271,6 @@ def is_day(utc_dt):
 @click.option('--min-cluster-size', help='The minimum number of samples in a group for that group to be considered a cluster. Default is 2. Increase for less conservative clustering, e.g. 5, 15', type=int)
 def run_cluster_roi(roi_dir, save_dir, device, config_ini, alpha, cluster_selection_epsilon, min_cluster_size):
     config = cfg.Config(config_ini)
-    max_area = int(config('cluster', 'max_area'))
-    min_area = int(config('cluster', 'min_area'))
     min_samples = int(config('cluster', 'min_samples'))
     alpha = alpha if alpha else float(config('cluster', 'alpha'))
     min_cluster_size = min_cluster_size if min_cluster_size else int(config('cluster', 'min_cluster_size'))

sdcat/cluster/embedding.py

@@ -11,6 +11,7 @@
 from numpy import save, load
 import numpy as np
 from sahi.utils.torch import torch
+from torchvision import transforms as pth_transforms
 import torch.nn as nn
 import cv2
 from sdcat.logger import info, err
@@ -119,8 +120,10 @@ def compute_embedding(images: list, model_name: str):
 
             image = np.array(square_img)
 
-            # Convert the image to a tensor
+            norm_transform = pth_transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
             img_tensor = torch.from_numpy(image).permute(2, 0, 1).float() / 255.0
+            # Noramlize the tensor with the mean and std of the ImageNet dataset
+            img_tensor = norm_transform(img_tensor)
             img_tensor = img_tensor.unsqueeze(0)  # Add batch dimension
             if 'cuda' in device:
                 img_tensor = img_tensor.to(device)