File size: 8,369 Bytes
b213d84
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
# Copyright (c) Facebook, Inc. and its affiliates.
import glob
import logging
import numpy as np
import os
import tempfile
from collections import OrderedDict
import torch
from PIL import Image

from detectron2.data import MetadataCatalog
from detectron2.utils import comm
from detectron2.utils.file_io import PathManager

from .evaluator import DatasetEvaluator


class CityscapesEvaluator(DatasetEvaluator):
    """
    Base class for evaluation using cityscapes API.
    """

    def __init__(self, dataset_name):
        """
        Args:
            dataset_name (str): the name of the dataset.
                It must have the following metadata associated with it:
                "thing_classes", "gt_dir".
        """
        self._metadata = MetadataCatalog.get(dataset_name)
        self._cpu_device = torch.device("cpu")
        self._logger = logging.getLogger(__name__)

    def reset(self):
        self._working_dir = tempfile.TemporaryDirectory(prefix="cityscapes_eval_")
        self._temp_dir = self._working_dir.name
        # All workers will write to the same results directory
        # TODO this does not work in distributed training
        assert (
            comm.get_local_size() == comm.get_world_size()
        ), "CityscapesEvaluator currently do not work with multiple machines."
        self._temp_dir = comm.all_gather(self._temp_dir)[0]
        if self._temp_dir != self._working_dir.name:
            self._working_dir.cleanup()
        self._logger.info(
            "Writing cityscapes results to temporary directory {} ...".format(self._temp_dir)
        )


class CityscapesInstanceEvaluator(CityscapesEvaluator):
    """
    Evaluate instance segmentation results on cityscapes dataset using cityscapes API.

    Note:
        * It does not work in multi-machine distributed training.
        * It contains a synchronization, therefore has to be used on all ranks.
        * Only the main process runs evaluation.
    """

    def process(self, inputs, outputs):
        from cityscapesscripts.helpers.labels import name2label

        for input, output in zip(inputs, outputs):
            file_name = input["file_name"]
            basename = os.path.splitext(os.path.basename(file_name))[0]
            pred_txt = os.path.join(self._temp_dir, basename + "_pred.txt")

            if "instances" in output:
                output = output["instances"].to(self._cpu_device)
                num_instances = len(output)
                with open(pred_txt, "w") as fout:
                    for i in range(num_instances):
                        pred_class = output.pred_classes[i]
                        classes = self._metadata.thing_classes[pred_class]
                        class_id = name2label[classes].id
                        score = output.scores[i]
                        mask = output.pred_masks[i].numpy().astype("uint8")
                        png_filename = os.path.join(
                            self._temp_dir, basename + "_{}_{}.png".format(i, classes)
                        )

                        Image.fromarray(mask * 255).save(png_filename)
                        fout.write(
                            "{} {} {}\n".format(os.path.basename(png_filename), class_id, score)
                        )
            else:
                # Cityscapes requires a prediction file for every ground truth image.
                with open(pred_txt, "w") as fout:
                    pass

    def evaluate(self):
        """
        Returns:
            dict: has a key "segm", whose value is a dict of "AP" and "AP50".
        """
        comm.synchronize()
        if comm.get_rank() > 0:
            return
        import cityscapesscripts.evaluation.evalInstanceLevelSemanticLabeling as cityscapes_eval

        self._logger.info("Evaluating results under {} ...".format(self._temp_dir))

        # set some global states in cityscapes evaluation API, before evaluating
        cityscapes_eval.args.predictionPath = os.path.abspath(self._temp_dir)
        cityscapes_eval.args.predictionWalk = None
        cityscapes_eval.args.JSONOutput = False
        cityscapes_eval.args.colorized = False
        cityscapes_eval.args.gtInstancesFile = os.path.join(self._temp_dir, "gtInstances.json")

        # These lines are adopted from
        # https://github.com/mcordts/cityscapesScripts/blob/master/cityscapesscripts/evaluation/evalInstanceLevelSemanticLabeling.py # noqa
        gt_dir = PathManager.get_local_path(self._metadata.gt_dir)
        groundTruthImgList = glob.glob(os.path.join(gt_dir, "*", "*_gtFine_instanceIds.png"))
        assert len(
            groundTruthImgList
        ), "Cannot find any ground truth images to use for evaluation. Searched for: {}".format(
            cityscapes_eval.args.groundTruthSearch
        )
        predictionImgList = []
        for gt in groundTruthImgList:
            predictionImgList.append(cityscapes_eval.getPrediction(gt, cityscapes_eval.args))
        results = cityscapes_eval.evaluateImgLists(
            predictionImgList, groundTruthImgList, cityscapes_eval.args
        )["averages"]

        ret = OrderedDict()
        ret["segm"] = {"AP": results["allAp"] * 100, "AP50": results["allAp50%"] * 100}
        self._working_dir.cleanup()
        return ret


class CityscapesSemSegEvaluator(CityscapesEvaluator):
    """
    Evaluate semantic segmentation results on cityscapes dataset using cityscapes API.

    Note:
        * It does not work in multi-machine distributed training.
        * It contains a synchronization, therefore has to be used on all ranks.
        * Only the main process runs evaluation.
    """

    def process(self, inputs, outputs):
        from cityscapesscripts.helpers.labels import trainId2label

        for input, output in zip(inputs, outputs):
            file_name = input["file_name"]
            basename = os.path.splitext(os.path.basename(file_name))[0]
            pred_filename = os.path.join(self._temp_dir, basename + "_pred.png")

            output = output["sem_seg"].argmax(dim=0).to(self._cpu_device).numpy()
            pred = 255 * np.ones(output.shape, dtype=np.uint8)
            for train_id, label in trainId2label.items():
                if label.ignoreInEval:
                    continue
                pred[output == train_id] = label.id
            Image.fromarray(pred).save(pred_filename)

    def evaluate(self):
        comm.synchronize()
        if comm.get_rank() > 0:
            return
        # Load the Cityscapes eval script *after* setting the required env var,
        # since the script reads CITYSCAPES_DATASET into global variables at load time.
        import cityscapesscripts.evaluation.evalPixelLevelSemanticLabeling as cityscapes_eval

        self._logger.info("Evaluating results under {} ...".format(self._temp_dir))

        # set some global states in cityscapes evaluation API, before evaluating
        cityscapes_eval.args.predictionPath = os.path.abspath(self._temp_dir)
        cityscapes_eval.args.predictionWalk = None
        cityscapes_eval.args.JSONOutput = False
        cityscapes_eval.args.colorized = False

        # These lines are adopted from
        # https://github.com/mcordts/cityscapesScripts/blob/master/cityscapesscripts/evaluation/evalPixelLevelSemanticLabeling.py # noqa
        gt_dir = PathManager.get_local_path(self._metadata.gt_dir)
        groundTruthImgList = glob.glob(os.path.join(gt_dir, "*", "*_gtFine_labelIds.png"))
        assert len(
            groundTruthImgList
        ), "Cannot find any ground truth images to use for evaluation. Searched for: {}".format(
            cityscapes_eval.args.groundTruthSearch
        )
        predictionImgList = []
        for gt in groundTruthImgList:
            predictionImgList.append(cityscapes_eval.getPrediction(cityscapes_eval.args, gt))
        results = cityscapes_eval.evaluateImgLists(
            predictionImgList, groundTruthImgList, cityscapes_eval.args
        )
        ret = OrderedDict()
        ret["sem_seg"] = {
            "IoU": 100.0 * results["averageScoreClasses"],
            "iIoU": 100.0 * results["averageScoreInstClasses"],
            "IoU_sup": 100.0 * results["averageScoreCategories"],
            "iIoU_sup": 100.0 * results["averageScoreInstCategories"],
        }
        self._working_dir.cleanup()
        return ret