File size: 3,761 Bytes
716e4bc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
# visuals.py
import torch
import matplotlib.pyplot as plt
import numpy as np
import gradio as gr
from PIL import Image

from uei_core.models import ModelPortfolio
from uei_core.uncertainty import UncertaintyEstimator
from uei_core.energy import EnergyProfiler

device = "cpu"

models = ModelPortfolio(device=device)
unc = UncertaintyEstimator()
energy = EnergyProfiler()


# ------------------------------
# 1️⃣ Plot: Uncertainty vs Energy Curve
# ------------------------------
def plot_unc_energy(img):
    x = models.preprocess(img)

    logits_s, E_s = energy.measure(models.infer_small, x)
    logits_l, E_l = energy.measure(models.infer_large, x)

    U_s = float(unc.estimate(logits_s))
    U_l = float(unc.estimate(logits_l))

    # Create plot
    fig, ax = plt.subplots(figsize=(5,4), dpi=120)

    xs = [E_s, E_l]
    ys = [U_s, U_l]
    labels = ["Small Model", "Large Model"]
    colors = ["#1f77b4", "#ff7f0e"]

    ax.scatter(xs, ys, s=150, color=colors)
    ax.plot(xs, ys, linestyle="--", color="#888")

    for i, label in enumerate(labels):
        ax.annotate(label, (xs[i], ys[i]), textcoords="offset points",
                    xytext=(8,5), ha='left', fontsize=10)

    ax.set_xlabel("Energy (proxy units)")
    ax.set_ylabel("Estimated Uncertainty")
    ax.set_title("Uncertainty vs Energy")
    ax.grid(True, alpha=0.3)

    return fig


# ------------------------------
# 2️⃣ Plot: Layer Activation Heatmap
# ------------------------------
def activation_heatmap(img):
    x = models.preprocess(img)

    # Register forward hook on the first conv
    activations = {}

    def hook(module, input, output):
        activations["feat"] = output.detach().cpu()

    h = models.small.features[0].register_forward_hook(hook)
    models.small(x)
    h.remove()

    feat = activations["feat"][0]  # first batch

    # Average channels → 2D heatmap
    heat = feat.mean(dim=0).numpy()

    fig, ax = plt.subplots(figsize=(4,4), dpi=120)
    ax.imshow(heat, cmap="viridis")
    ax.set_title("Early Layer Activation Heatmap")
    ax.axis("off")

    return fig


# ------------------------------
# 3️⃣ Plot: Model Comparison Bars
# ------------------------------
def model_comparison(img):
    x = models.preprocess(img)

    logits_s, E_s = energy.measure(models.infer_small, x)
    logits_l, E_l = energy.measure(models.infer_large, x)

    U_s = float(unc.estimate(logits_s))
    U_l = float(unc.estimate(logits_l))

    fig, ax = plt.subplots(figsize=(6,4))

    labels = ["Small Model", "Large Model"]
    energy_vals = [E_s, E_l]
    unc_vals = [U_s, U_l]

    x_axis = np.arange(len(labels))
    w = 0.35

    ax.bar(x_axis - w/2, energy_vals, w, label="Energy", color="#2ca02c")
    ax.bar(x_axis + w/2, unc_vals, w, label="Uncertainty", color="#d62728")

    ax.set_xticks(x_axis)
    ax.set_xticklabels(labels)
    ax.set_title("Model Energy & Uncertainty Comparison")
    ax.legend()
    ax.grid(alpha=0.2)

    return fig


# ------------------------------
#  🔥 Gradio Interface
# ------------------------------
def get_visual_ui():
    with gr.Blocks() as demo:
        gr.Markdown("## 🔍 UEI Visualization Dashboard")
        gr.Markdown("Explore how UEI behaves internally with colorful charts")

        img = gr.Image(type="pil", label="Upload Image")

        with gr.Tabs():
            with gr.Tab("Uncertainty vs Energy"):
                gr.Plot(label="Chart").render(fn=plot_unc_energy, inputs=img)

            with gr.Tab("Layer Activations"):
                gr.Plot(label="Activation Heatmap").render(fn=activation_heatmap, inputs=img)

            with gr.Tab("Model Comparison"):
                gr.Plot(label="Energy & Uncertainty Bars").render(fn=model_comparison, inputs=img)

    return demo