LearnableEarthParser/test.py at main · cnstt/LearnableEarthParser · GitHub

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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
import torch
import spconv.pytorch as spconv

import logging
import yaml
import hydra
import os
import os.path as osp
import pytorch_lightning as pl
from omegaconf import DictConfig, OmegaConf

import matplotlib.pyplot as plt
plt.switch_backend('agg')

import multidti3d

logger = logging.getLogger(__name__)

@hydra.main(config_path="configs", config_name="test")
def main(cfg: DictConfig) -> None:

    html = f"""
    <html>
    <head>
    <style>
    table {{
    font-family: arial, sans-serif;
    border-collapse: collapse;
    width: 100%;
    }}
    td, th {{
    border: 1px solid #dddddd;
    text-align: left;
    padding: 8px;
    }}
    tr:nth-child(even) {{
    background-color: #dddddd;
    }}
    </style>
    </head>
    <body>
    <h1>Results</h1>
    <table>
    <tr>
        <th>Experiment</th>
        <th>Chamfer</th>
        <th>mIoU</th>
        <th>Acc</th>
        <th>RGB</th>
        <th>Intensity</th>
        <th>Ground truth</th>
        <th>Prediction</th>
        <th>Instance pred</th>
        <th>Rec intensity</th>
        <th>Rec prediction</th>
        <th>Protos</th>
        <th>Protos labels</th>
        <th>Report</th>
    </tr>
    """

    root = cfg.test.root

    for xp in sorted(os.listdir(root)):
        if osp.exists(osp.join(root, xp, "checkpoints")):
            logger.info("")
            logger.info(f"{xp}")

            root_xp = osp.join(root, xp)
            ckpts = osp.join(root_xp, "checkpoints")

            epoch = 0
            for ckpt in os.listdir(ckpts):
                if "step" in ckpt and ckpt.endswith(".ckpt"):
                    this_epoch = int(ckpt.split("=")[-1].split(".")[0])
                    if this_epoch > epoch:
                        epoch = this_epoch
                        last_ckpt = osp.join(ckpts, ckpt)

            overrides = osp.join(root_xp, ".hydra", "overrides.yaml")
            with open(overrides, "r") as stream:
                overrides = yaml.safe_load(stream)

            cfg = hydra.compose(
                config_name="defaults",
                overrides=overrides + [
                    "hydra.searchpath=[file://../../../EarthParserDataset/configs]",
                    f"model.callbacks.outhtml.points_per_scene=0",
                    f"mode=test",
                ]
            )

            cfg.model.load_weights = last_ckpt

            logger.info(f"{xp}\tcfg loaded : {last_ckpt}")

            if cfg.seed != 0: pl.seed_everything(cfg.seed)

            cfg.data._target_ = f"EarthParserDataset.{cfg.data._target_}"
            datamodule = hydra.utils.instantiate(cfg.data)
            datamodule.setup("train")
            datamodule.setup("test")

            logger.info(f"{xp}\tDatamodule loaded : {datamodule}")

            model = hydra.utils.instantiate(
                cfg.model,
                _recursive_=False,
            )
            model.__initprotos__(datamodule)

            if cfg.model.load_weights != "":
                logger.info(f"{xp}\tLoading weights from {cfg.model.load_weights}")
                model.load_state_dict(torch.load(cfg.model.load_weights)["state_dict"], strict=False)

            trainer = multidti3d.trainers.get_trainer(cfg)

            trainer.test(model, datamodule=datamodule)

            if "fit" in cfg.model.load_weights:
                modelname = model.hparams.load_weights.split("fit/")[-1].split("/")[0]
            else:
                modelname = model.hparams.load_weights.split("outputs/")[-1].split("/")[0]

            html += f"""
            <tr>
                <td>{cfg.experiment}</td>
                <td>{1000*trainer.logged_metrics['Losses/chamfer/test']:.3f}</td>
                <td>{100*trainer.logged_metrics['IoU/test']:.3f}</td>
                <td>{100*trainer.logged_metrics['Acc/test']:.3f}</td>
                <td><img width=100px src="{cfg.data.name}_{modelname}_input_rgb_0.png"/></td>
                <td><img width=100px src="{cfg.data.name}_{modelname}_input_intensity_0.png"/></td>
                <td><img width=100px src="{cfg.data.name}_{modelname}_input_point_y_0.png"/></td>
                <td><img width=100px src="{cfg.data.name}_{modelname}_input_point_y_pred_0.png"/></td>
                <td><img width=100px src="{cfg.data.name}_{modelname}_input_point_inst_pred_0.png"/></td>
                <td><img width=100px src="{cfg.data.name}_{modelname}_reconstruction_intensity_0.png"/></td>
                <td><img width=100px src="{cfg.data.name}_{modelname}_reconstruction_point_y_0.png"/></td>
                <td><img width=100px src="{cfg.data.name}_{modelname}_prototypes_rgb_0.png"/></td>
                <td><img width=100px src="{cfg.data.name}_{modelname}_prototypes_point_y_0.png"/></td>
                <td><a href="report_{cfg.data.name}_{modelname}_on_test_end.html">report_{cfg.data.name}_{modelname}_on_test_end.html</a></td>
            </tr>
            """

            logger.info("{xp}\tDone !")
            del trainer, model, datamodule, cfg

    html += """
    </body>
    </html>
    """

    with open(f"index.html", "w") as f:
        f.write(html)

if __name__ == '__main__':
    main()