X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=world.py;h=fb5d5c796990ef476ff79fb9ead66615a17fd7fd;hb=a8f039a9b491b1b4b47f6b9f8123c7261e758661;hp=bac9e761e248bb64547aada2bac7109f0099d38d;hpb=e38b98574f1966ea3a91ffb8fd9042f10a75ca88;p=picoclvr.git

diff --git a/world.py b/world.py
index bac9e76..fb5d5c7 100755
--- a/world.py
+++ b/world.py
@@ -30,7 +30,7 @@ class Box:
         return False
 
 
-def scene2tensor(xh, yh, scene, size=512):
+def scene2tensor(xh, yh, scene, size=64):
     width, height = size, size
     pixel_map = torch.ByteTensor(width, height, 4).fill_(255)
     data = pixel_map.numpy()
@@ -48,11 +48,9 @@ def scene2tensor(xh, yh, scene, size=512):
         ctx.rel_line_to(-b.w * size, 0)
         ctx.close_path()
         ctx.set_source_rgba(b.r, b.g, b.b, 1.0)
-        ctx.fill_preserve()
-        ctx.set_source_rgba(0, 0, 0, 1.0)
-        ctx.stroke()
+        ctx.fill()
 
-    hs = size * 0.05
+    hs = size * 0.1
     ctx.set_source_rgba(0.0, 0.0, 0.0, 1.0)
     ctx.move_to(xh * size - hs / 2, yh * size - hs / 2)
     ctx.rel_line_to(hs, 0)
@@ -69,7 +67,7 @@ def random_scene():
     colors = [
         (1.00, 0.00, 0.00),
         (0.00, 1.00, 0.00),
-        (0.00, 0.00, 1.00),
+        (0.60, 0.60, 1.00),
         (1.00, 1.00, 0.00),
         (0.75, 0.75, 0.75),
     ]
@@ -87,7 +85,7 @@ def random_scene():
     return scene
 
 
-def sequence(length=10):
+def sequence(nb_steps=10, all_frames=False):
     delta = 0.1
     effects = [
         (False, 0, 0),
@@ -102,12 +100,14 @@ def sequence(length=10):
     ]
 
     while True:
+        frames = []
+
         scene = random_scene()
         xh, yh = tuple(x.item() for x in torch.rand(2))
 
-        frame_start = scene2tensor(xh, yh, scene)
+        frames.append(scene2tensor(xh, yh, scene))
 
-        actions = torch.randint(len(effects), (length,))
+        actions = torch.randint(len(effects), (nb_steps,))
         change = False
 
         for a in actions:
@@ -137,14 +137,123 @@ def sequence(length=10):
                 if xh < 0 or xh > 1 or yh < 0 or yh > 1:
                     xh, yh = x, y
 
-        frame_end = scene2tensor(xh, yh, scene)
+            if all_frames:
+                frames.append(scene2tensor(xh, yh, scene))
+
+        if not all_frames:
+            frames.append(scene2tensor(xh, yh, scene))
+
         if change:
             break
 
-    return frame_start, frame_end, actions
+    return frames, actions
+
+
+######################################################################
+
+
+# ||x_i - c_j||^2 = ||x_i||^2 + ||c_j||^2 - 2<x_i, c_j>
+def sq2matrix(x, c):
+    nx = x.pow(2).sum(1)
+    nc = c.pow(2).sum(1)
+    return nx[:, None] + nc[None, :] - 2 * x @ c.t()
+
+
+def update_centroids(x, c, nb_min=1):
+    _, b = sq2matrix(x, c).min(1)
+    b.squeeze_()
+    nb_resets = 0
+
+    for k in range(0, c.size(0)):
+        i = b.eq(k).nonzero(as_tuple=False).squeeze()
+        if i.numel() >= nb_min:
+            c[k] = x.index_select(0, i).mean(0)
+        else:
+            n = torch.randint(x.size(0), (1,))
+            nb_resets += 1
+            c[k] = x[n]
+
+    return c, b, nb_resets
+
+
+def kmeans(x, nb_centroids, nb_min=1):
+    if x.size(0) < nb_centroids * nb_min:
+        print("Not enough points!")
+        exit(1)
+
+    c = x[torch.randperm(x.size(0))[:nb_centroids]]
+    t = torch.full((x.size(0),), -1)
+    n = 0
+
+    while True:
+        c, u, nb_resets = update_centroids(x, c, nb_min)
+        n = n + 1
+        nb_changes = (u - t).sign().abs().sum() + nb_resets
+        t = u
+        if nb_changes == 0:
+            break
+
+    return c, t
+
+
+######################################################################
+
+
+def patchify(x, factor, invert_size=None):
+    if invert_size is None:
+        return (
+            x.reshape(
+                x.size(0), #0
+                x.size(1), #1
+                factor, #2
+                x.size(2) // factor,#3
+                factor,#4
+                x.size(3) // factor,#5
+            )
+            .permute(0, 2, 4, 1, 3, 5)
+            .reshape(-1, x.size(1), x.size(2) // factor, x.size(3) // factor)
+        )
+    else:
+        return (
+            x.reshape(
+                invert_size[0], #0
+                factor, #1
+                factor, #2
+                invert_size[1], #3
+                invert_size[2] // factor, #4
+                invert_size[3] // factor, #5
+            )
+            .permute(0, 3, 1, 4, 2, 5)
+            .reshape(invert_size)
+        )
 
 
 if __name__ == "__main__":
-    frame_start, frame_end, actions = sequence()
-    torchvision.utils.save_image(frame_start, "world_start.png")
-    torchvision.utils.save_image(frame_end, "world_end.png")
+    import time
+
+    all_frames = []
+    nb = 1000
+    start_time = time.perf_counter()
+    for n in range(nb):
+        frames, actions = sequence(nb_steps=31)
+        all_frames += frames
+    end_time = time.perf_counter()
+    print(f"{nb / (end_time - start_time):.02f} samples per second")
+
+    input = torch.cat(all_frames, 0)
+    x = patchify(input, 8)
+    y = x.reshape(x.size(0), -1)
+    print(f"{x.size()=} {y.size()=}")
+    centroids, t = kmeans(y, 4096)
+    results = centroids[t]
+    results = results.reshape(x.size())
+    results = patchify(results, 8, input.size())
+
+    print(f"{input.size()=} {results.size()=}")
+
+    torchvision.utils.save_image(input[:64], "orig.png", nrow=8)
+    torchvision.utils.save_image(results[:64], "qtiz.png", nrow=8)
+
+    # frames, actions = sequence(nb_steps=31, all_frames=True)
+    # frames = torch.cat(frames, 0)
+    # torchvision.utils.save_image(frames, "seq.png", nrow=8)