File size: 2,061 Bytes
021d920 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 | """R-grid: trainless learnability ratio over size x contrast, both bgs.
Same R as compute_R.py (pooled object-toggle signal / temporal+spatial clutter),
just vectorized over a grid and dumped to JSON. Purpose: test whether R measured
AT THE TARGET CONTRAST already flags double-deficit (small AND low-contrast)
cells that compiler v1's single-axis router mis-handled.
"""
import json, numpy as np, cv2, sys
B = np.load("simreal/frames64.npy").astype(np.float32) / 255.0
rng = np.random.default_rng(0)
def pool(img): return cv2.resize(img, (8, 8), interpolation=cv2.INTER_AREA).ravel()
def draw(img, pos, c, size, ct):
img = img.copy(); x, y = pos
img[y:y+size, x:x+size] = c*ct + img[y:y+size, x:x+size].mean(axis=(0, 1))*(1-ct)
return img
def R(mode, size, ct, n=80):
sig, den = [], []
for _ in range(n):
if mode == "static": frames = [B[rng.integers(len(B))]]*9
else: frames = [B[rng.integers(len(B))] for _ in range(9)]
pos = (int(rng.integers(2, 60-size)), int(rng.integers(2, 60-size)))
c = rng.uniform(0.15, 0.95, 3).astype(np.float32)
sig.append(np.sum((pool(draw(frames[0], pos, c, size, ct))
- pool(draw(frames[0], pos, 1-c, size, ct)))**2)/4)
P = np.array([pool(f) for f in frames])
probes = [pool(np.roll(np.roll(frames[0], int(rng.integers(64)), 0),
int(rng.integers(64)), 1)) for _ in range(6)]
den.append(P.var(axis=0).sum() + np.array(probes).var(axis=0).sum())
return float(np.mean(sig)/(np.mean(den)+1e-9))
SIZES = [5, 6, 8, 10, 12, 14, 16]
CTS = [1.0, 0.4]
TH = 0.02
grid = {}
for mode in ["static", "moving"]:
for ct in CTS:
for s in SIZES:
r = R(mode, s, ct)
grid[f"{mode}_o{s}_ct{ct:g}"] = round(r, 4)
flag = "DIRECT" if r >= TH else "floor"
print(f"{mode:6s} {s:2d}px ct{ct:<3g} R={r:.4f} {flag}", flush=True)
json.dump(grid, open("results_compiler_v2/R_grid.json", "w"), indent=2)
print("\nwrote results_compiler_v2/R_grid.json")
|