[ADD] Random Embedder

app.py

@@ -6,8 +6,7 @@ import plotly.graph_objects as go
 
 from geoguessr_bot.guessr import RandomGuessr, AbstractGuessr, NearestNeighborEmbedderGuessr, \
     AverageNeighborsEmbedderGuessr
-from geoguessr_bot.retriever import DinoV2Embedder, Retriever
-
+from geoguessr_bot.retriever import DinoV2Embedder, Retriever, RandomEmbedder
 
 ALL_GUESSR_CLASS = {
     "random": RandomGuessr,
@@ -21,6 +20,7 @@ ALL_GUESSR_ARGS = {
         "embedder": DinoV2Embedder(
             device="cpu"
         ),
+        # "embedder": RandomEmbedder(n_dim=384),
         "retriever": Retriever(
             embeddings_path=os.path.join(os.path.dirname(os.path.abspath(__file__)),
                                          "resources/embeddings.npy"),
@@ -32,13 +32,14 @@ ALL_GUESSR_ARGS = {
         "embedder": DinoV2Embedder(
             device="cpu"
         ),
+        # "embedder": RandomEmbedder(n_dim=384),
         "retriever": Retriever(
             embeddings_path=os.path.join(os.path.dirname(os.path.abspath(__file__)),
                                          "resources/embeddings.npy"),
         ),
         "metadata_path": os.path.join(os.path.dirname(os.path.abspath(__file__)),
                                         "resources/metadatav3.csv"),
-        "n_neighbors": 2000,
+        "n_neighbors": 100,
         "dbscan_eps": 0.5
     }
 }

geoguessr_bot/guessr/average_neighbor_embedder_guessr.py

@@ -13,6 +13,11 @@ from geoguessr_bot.retriever import AbstractImageEmbedder
 from geoguessr_bot.retriever import Retriever
 
 
+def haversine_distance(x, y) -> float:
+    """Compute the haversine distance between two coordinates
+    """
+    return haversine_distances(np.array(x).reshape(1, -1), np.array(y).reshape(1, -1))[0][0]
+
 @dataclass
 class AverageNeighborsEmbedderGuessr(AbstractGuessr):
     """Guesses a coordinate using an Embedder and a retriever followed by NN.
@@ -20,7 +25,7 @@ class AverageNeighborsEmbedderGuessr(AbstractGuessr):
     embedder: AbstractImageEmbedder
     retriever: Retriever
     metadata_path: str
-    n_neighbors: int = 1000
+    n_neighbors: int = 50
     dbscan_eps: float = 0.05
 
     def __post_init__(self):
@@ -32,7 +37,7 @@ class AverageNeighborsEmbedderGuessr(AbstractGuessr):
             for image, latitude, longitude in zip(metadata["path"], metadata["latitude"], metadata["longitude"])
         }
         # DBSCAN will be used to take the centroid of the biggest cluster among the N neighbors, using Haversine
-        self.dbscan = DBSCAN(eps=self.dbscan_eps, metric=haversine_distances)
+        self.dbscan = DBSCAN(eps=self.dbscan_eps, metric=haversine_distance)
 
     def guess(self, image: Image) -> Coordinate:
         """Guess a coordinate from an image

geoguessr_bot/retriever/__init__.py

@@ -2,3 +2,4 @@ from .retriever import Retriever
 from .abstract_embedder import AbstractImageEmbedder
 from .dino_embedder import DinoEmbedder
 from .dino_v2_embedder import DinoV2Embedder
+from .random_embedder import RandomEmbedder

geoguessr_bot/retriever/abstract_embedder.py

@@ -1,4 +1,6 @@
 import os
+from abc import abstractmethod
+
 from PIL import Image
 import numpy as np
 from tqdm import tqdm
@@ -8,6 +10,7 @@ class AbstractImageEmbedder:
     def __init__(self, device: str = "cpu"):
         self.device = device
 
+    @abstractmethod
     def embed(self, image: Image) -> np.ndarray:
         """Embed an image
         """

geoguessr_bot/retriever/random_embedder.py

@@ -0,0 +1,16 @@
+from dataclasses import dataclass
+
+import numpy as np
+from PIL import Image
+
+from geoguessr_bot.retriever import AbstractImageEmbedder
+
+
+@dataclass
+class RandomEmbedder(AbstractImageEmbedder):
+    n_dim: int = 8
+
+    def embed(self, image: Image) -> np.ndarray:
+        """Embed an image
+        """
+        return np.random.rand(self.n_dim)