From c27bceb6d410ece45e0f1b723b761e004fe9c7b2 Mon Sep 17 00:00:00 2001
From: James Ketrenos <james_gitlab@ketrenos.com>
Date: Sat, 11 Jan 2020 19:38:23 -0800
Subject: [PATCH] Recursively reduce cluster sets

Signed-off-by: James Ketrenos <james_gitlab@ketrenos.com>
---
 scanner/scanner.c | 34 +++++++++++++++++++++++++++++-----
 1 file changed, 29 insertions(+), 5 deletions(-)

diff --git a/scanner/scanner.c b/scanner/scanner.c
index 06598e5..cce8fd8 100644
--- a/scanner/scanner.c
+++ b/scanner/scanner.c
@@ -14,6 +14,10 @@
 #define MAX_DISTANCE 0.47L
 #endif
 
+#ifndef MAX_CLUSTER_SIZE
+#define MAX_CLUSTER_SIZE 150
+#endif
+
 /* descriptors extracted from a profile face photo, used to filter out faces near a profile image 
  * (which all identify as the same person) */
 float profileDescriptors[][128] = {
@@ -315,6 +319,7 @@ int main(int argc, char *argv[]) {
   long int entries = 0;
   long int minPts = MIN_PTS;
   float maxDistance = MAX_DISTANCE;
+  long int maxClusterSize = MAX_CLUSTER_SIZE;
   
   if (argc == 1) {
     fprintf(stderr, "usage: scanner PATH MAX_DISTANCE MIN_PTS\n");
@@ -329,6 +334,10 @@ int main(int argc, char *argv[]) {
     sscanf(argv[3], "%ld", &minPts);
   }
 
+  if (argv > 4) {
+    sscanf(argv[4], "%ld", &maxClusterSize);
+  }
+
   fprintf(stderr, "\nmaxDistance : %f\nminPts      : %ld\n", maxDistance, minPts);
 
   /* Allocate storage for all distances */
@@ -475,13 +484,22 @@ int main(int argc, char *argv[]) {
   long int clusters = DBSCAN(ppFaces, entries, maxDistance, minPts, -1);
   fprintf(stderr, "\n%ld clusters identified before size-split.\n", clusters);
 
-  if (clusters > 0) {
+  int recalcNeeded = clusters > 0 ? 1 : 0;
+  float reducedDistance = maxDistance;
+
+  while (recalcNeeded) {
     int *stats = malloc(sizeof(int) * clusters), delta = 0;
     memset(stats, 0, sizeof(int) * clusters);
     getClusterCounts(stats, ppFaces, entries);
-    
+    recalcNeeded = 0;
+
+    reducedDistance -= 0.05L;
+    if (reducedDistance < 0.1) {
+      break;
+    }
+
     for (int i = 0; i < clusters; i++) {
-      if (stats[i] < 100) {
+      if (stats[i] < maxClusterSize) {
         continue;
       }
 
@@ -492,8 +510,14 @@ int main(int argc, char *argv[]) {
         }
       }
 
-      int split = DBSCAN(ppFaces, entries, maxDistance - 0.05L, minPts, i + 1);
-      fprintf(stderr, "Cluster %d had %d units. Split into %d clusters.\n", i + 1, stats[i], split);
+      int split = DBSCAN(ppFaces, entries, reducedDistance, minPts, i + 1);
+      if (split) {
+        recalcNeeded |= 1;
+      } else {
+        continue;
+      }
+      fprintf(stderr, "Cluster %d had %d units. Split into %d clusters (max: %f).\n",
+          i + 1, stats[i], split, reducedDistance);
       for (int c = 0; c < split; c++) {
         fprintf(stderr, "%ld. %ld\n", c + clusters, getClusterCount(ppFaces, entries, delta + c + clusters));
       }