Added profile distance back in to scanner

Signed-off-by: James Ketrenos <james_gitlab@ketrenos.com>

scanner/scanner.c

@@ -36,6 +36,7 @@ typedef struct Face {
   long faceId;
   long photoId;
   double confidence;
+  double profileDistance;
   ClusterTypes clusterType;
   double *distances;
 } Face;
@@ -46,45 +47,6 @@ typedef struct FaceLink {
   Face *pFace;
 } FaceLink;
 
-char fileBuf[5000];
-char pathBuf[1028];
-
-
-Face *readFaceDescriptor(Face *pFace, long id, char *path) {
-  FILE *f;
-  f = fopen(path, "r");
-  if (!f) {
-    return NULL;
-  }
-  size_t s = fread(fileBuf, 1, sizeof(fileBuf), f);
-  fclose(f);
-  
-  char *p = fileBuf;
-  fileBuf[s] = 0;
-  while (*p && *p != '-' && *p != '+' && (*p < '0' || *p > '9')) {
-    p++;
-  }
-  int i = 0;
-  for (i = 0; i < 128; i++) {
-    char *start = p;
-    while (*p && *p != ',' && *p != ']' && *p != ' ' && *p != '\n') {
-      p++;
-    }
-    if (!*p) {
-      break;
-    }
-    *p++ = 0;
-    sscanf(start, "%Lf", &pFace->descriptor[i]);
-  }
-
-  if (i != 128) {
-    return NULL;
-  }
-
-  pFace->faceId = id;
-
-  return pFace;
-}
 
 long double euclideanDistance(long double *a, long double *b) {
   long double sum = 0.0L;
@@ -136,7 +98,7 @@ FaceLink *RangeQuery(Face **ppFaces, long int faceCount, Face *pQ, double eps) {
   FaceLink *pNeighbors = NULL;
   for (long int i = 0; i < faceCount; i++) {
     Face *pFace = ppFaces[i];
-    if (pFace->confidence <= 0.9) {
+    if (pFace->profileDistance <= 0.5) {
       continue;
     }
 
@@ -177,7 +139,7 @@ long int DBSCAN(Face **ppFaces, long int faceCount, double eps, int minPts) {
   long int C = 0;
   for (long int i = 0; i < faceCount; i++) {
     Face *pFace = ppFaces[i];
-    if (pFace->confidence <= 0.9) {
+    if (pFace->profileDistance <= 0.5) {
       continue;
     }
 
@@ -318,10 +280,15 @@ int parseFaceIdRow(void *data, int argc, char **argv, char **column) {
   long int photoId = strtol(argv[1] ? argv[1] : "0", NULL, 10);
   double confidence = strtod(argv[2] ? argv[2] : "0.0", NULL);
 
+  if (confidence < 0.9) {
+    return 0;
+  }
+
   Face *pFace = map->ppFaces[map->initialized++];
   pFace->faceId = faceId;
   pFace->photoId = photoId;
   pFace->confidence = confidence;
+
   return 0;
 }
 
@@ -414,6 +381,8 @@ int main(int argc, char *argv[]) {
     sqlite3_close(db);
     return 1;
   }
+  entries = data.initialized;
+  data.count = data.initialized;
 
   fprintf(stderr, "Face data loaded from DB.\n");
 
@@ -427,11 +396,33 @@ int main(int argc, char *argv[]) {
 
   fprintf(stderr, "Descriptor data loaded from DB\n");
 
-  err_msg = NULL;
+  long double profileDistance = 1.0;
+  long int dst, src;
+  for (src = 0, dst = 0; src < entries; src++) {
+    Face *pFace = ppFaces[src];
+    profileDistance = 1.0;
+    for (int j = 0; j < (sizeof(profileDescriptors) / sizeof(profileDescriptors[0])); j++) {
+      profileDistance = euclideanDistance(pFace->descriptor, profileDescriptors[j]);
+      if (profileDistance > 0.5) {
+        profileDistance = 1.0;
+      } else {
+        break;
+      }
+    }
+    if (profileDistance <= 0.5) {
+      free(pFace->distances);
+    } else {
+      pFace->profileDistance = profileDistance;
+      ppFaces[dst++] = pFace;
+    }
+  }
+  fprintf(stderr, "Dropped %ld faces as too close to profile photos.\n", (entries - dst));
+  entries = dst;
 
   long int processed = 0;
   long int last = 0;
   long double total = 0.0;
+  long int sampleSize = 0;
 
   for (long i = 0; i < entries; i++) {
     Face *pLink = ppFaces[i];
@@ -446,12 +437,6 @@ int main(int argc, char *argv[]) {
         }
       }
 
-      if (pLink->confidence <= 0.9 || pTarget->confidence <= 0.9) {
-        pLink->distances[i] = 0.0;
-        pTarget->distances[j] = 0.0;
-        continue;
-      }
-
       if (i == j) {
         pLink->distances[i] = 0.0;
         pTarget->distances[j] = 0.0;
@@ -465,11 +450,12 @@ int main(int argc, char *argv[]) {
       pLink->distances[j] = 
         pTarget->distances[i] = euclideanDistance(pLink->descriptor, pTarget->descriptor);
 
+      sampleSize++;
       total += pLink->distances[j];
     }
   }
 
-  fprintf(stderr, "\nAverage distance: %Lf\n", 1. * total / (entries * entries));
+  fprintf(stderr, "\nAverage distance: %Lf\n", 1. * total / sampleSize);
 
   long int clusters = DBSCAN(ppFaces, entries, maxDistance, minPts);
   long int undefined = 0, outlier = 0, core = 0, reachable = 0;