[Feature]: support image visualization for tensorboard and wandb

benchmarks/extract_dataset_configs/imagenet.py

@@ -23,4 +23,5 @@
             dict(type='CenterCrop', size=224),
             dict(type='ToTensor'),
             dict(type='Normalize', **img_norm_cfg),
+            dict(type='Collect', keys=['img', 'gt_labels'])
         ]))

configs/detection/yolox/yolox_s_8xb16_300e_coco.py

@@ -130,7 +130,14 @@
 ]
 
 # evaluation
-eval_config = dict(interval=10, gpu_collect=False)
+eval_config = dict(
+    interval=10,
+    gpu_collect=False,
+    visualization_config=dict(
+        vis_num=10,
+        score_thr=0.5,
+    )  # show by TensorboardLoggerHookV2 and WandbLoggerHookV2
+)
 eval_pipelines = [
     dict(
         mode='test',
@@ -168,7 +175,8 @@
     interval=100,
     hooks=[
         dict(type='TextLoggerHook'),
-        dict(type='TensorboardLoggerHook')
+        dict(type='TensorboardLoggerHookV2'),
+        # dict(type='WandbLoggerHookV2'),
     ])
 
 export = dict(use_jit=False)

easycv/core/visualization/__init__.py

@@ -1,4 +1,4 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
-from .image import imshow_bboxes, imshow_keypoints
+from .image import imshow_bboxes, imshow_keypoints, imshow_label
 
-__all__ = ['imshow_bboxes', 'imshow_keypoints']
+__all__ = ['imshow_bboxes', 'imshow_keypoints', 'imshow_label']

easycv/core/visualization/image.py

@@ -1,18 +1,119 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 # Adapt from https://github.com/open-mmlab/mmpose/blob/master/mmpose/core/visualization/image.py
 import math
+import os
+from urllib import request
 
 import cv2
 import mmcv
 import numpy as np
 from mmcv.utils.misc import deprecated_api_warning
+from PIL import Image, ImageDraw, ImageFont
+from torch.hub import DEFAULT_CACHE_DIR
+
+
+def download_font(save_path=None):
+    url_path = 'http://pai-vision-data-hz.oss-accelerate.aliyuncs.com/EasyCV/pkgs/simhei.ttf'
+    if save_path is None:
+        save_dir = DEFAULT_CACHE_DIR
+        if not os.path.exists(save_dir):
+            os.makedirs(save_dir)
+        save_path = os.path.join(save_dir, 'simhei.ttf')
+
+    if os.path.exists(save_path):
+        return save_path
+
+    f = request.urlopen(url_path)
+    with open(save_path, 'wb') as fw:
+        fw.write(f.read())
+
+    return save_path
+
+
+def put_text(img, xy, text, fill, font_path, size=20):
+    """support chinese text
+    """
+    img = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
+    draw = ImageDraw.Draw(img)
+    fontText = ImageFont.truetype(font_path, size=size, encoding='utf-8')
+    draw.text(xy, text, fill=fill, font=fontText)
+    img = cv2.cvtColor(np.asarray(img), cv2.COLOR_RGB2BGR)
+    return img
+
+
+def imshow_label(img,
+                 labels,
+                 text_color='blue',
+                 font_size=20,
+                 thickness=1,
+                 font_scale=0.5,
+                 intervel=5,
+                 show=True,
+                 win_name='',
+                 wait_time=0,
+                 out_file=None):
+    """Draw images with labels on an image.
+
+    Args:
+        img (str or ndarray): The image to be displayed.
+        labels (str or list[str]): labels of each image.
+        text_color (str or tuple or :obj:`Color`): Color of texts.
+        font_size (int): Size of font.
+        thickness (int): Thickness of lines.
+        font_scale (float): Font scales of texts.
+        intervel（int): interval pixels between multiple labels
+        show (bool): Whether to show the image.
+        win_name (str): The window name.
+        wait_time (int): Value of waitKey param.
+        out_file (str, optional): The filename to write the image.
+
+    Returns:
+        ndarray: The image with bboxes drawn on it.
+    """
+    font_path = download_font()
+    img = mmcv.imread(img)
+    img = np.ascontiguousarray(img)
+    labels = [labels] if isinstance(labels, str) else labels
+
+    cur_height = 0
+    for label in labels:
+        # roughly estimate the proper font size
+        text_size, text_baseline = cv2.getTextSize(label,
+                                                   cv2.FONT_HERSHEY_DUPLEX,
+                                                   font_scale, thickness)
+
+        org = (text_baseline + text_size[1],
+               text_baseline + text_size[1] + cur_height)
+
+        # support chinese text
+        # TODO: Unify the font of cv2 and PIL, and auto get font_size according to the font_scale
+        img = put_text(
+            img,
+            org,
+            text=label,
+            fill=text_color,
+            font_path=font_path,
+            size=font_size)
+
+        # cv2.putText(img, label, org, cv2.FONT_HERSHEY_DUPLEX, font_scale,
+        #             mmcv.color_val(text_color), thickness)
+
+        cur_height += text_baseline + text_size[1] + intervel
+
+    if show:
+        mmcv.imshow(img, win_name, wait_time)
+    if out_file is not None:
+        mmcv.imwrite(img, out_file)
+
+    return img
 
 
 def imshow_bboxes(img,
                   bboxes,
                   labels=None,
                   colors='green',
                   text_color='white',
+                  font_size=20,
                   thickness=1,
                   font_scale=0.5,
                   show=True,
@@ -29,6 +130,7 @@ def imshow_bboxes(img,
         labels (str or list[str], optional): labels of each bbox.
         colors (list[str or tuple or :obj:`Color`]): A list of colors.
         text_color (str or tuple or :obj:`Color`): Color of texts.
+        font_size (int): Size of font.
         thickness (int): Thickness of lines.
         font_scale (float): Font scales of texts.
         show (bool): Whether to show the image.
@@ -39,6 +141,7 @@ def imshow_bboxes(img,
     Returns:
         ndarray: The image with bboxes drawn on it.
     """
+    font_path = download_font()
 
     # adapt to mmcv.imshow_bboxes input format
     bboxes = np.split(
@@ -58,11 +161,10 @@ def imshow_bboxes(img,
         out_file=None)
 
     if labels is not None:
-        if not isinstance(labels, list):
-            labels = [labels for _ in range(len(bboxes))]
         assert len(labels) == len(bboxes)
 
         for bbox, label, color in zip(bboxes, labels, colors):
+            label = str(label)
             bbox_int = bbox[0, :4].astype(np.int32)
             # roughly estimate the proper font size
             text_size, text_baseline = cv2.getTextSize(label,
@@ -74,9 +176,18 @@ def imshow_bboxes(img,
             text_y2 = text_y1 + text_size[1] + text_baseline
             cv2.rectangle(img, (text_x1, text_y1), (text_x2, text_y2), color,
                           cv2.FILLED)
-            cv2.putText(img, label, (text_x1, text_y2 - text_baseline),
-                        cv2.FONT_HERSHEY_DUPLEX, font_scale,
-                        mmcv.color_val(text_color), thickness)
+            # cv2.putText(img, label, (text_x1, text_y2 - text_baseline),
+            #             cv2.FONT_HERSHEY_DUPLEX, font_scale,
+            #             mmcv.color_val(text_color), thickness)
+
+            # support chinese text
+            # TODO: Unify the font of cv2 and PIL, and auto get font_size according to the font_scale
+            img = put_text(
+                img, (text_x1, text_y1),
+                text=label,
+                fill=text_color,
+                font_path=font_path,
+                size=font_size)
 
     if show:
         mmcv.imshow(img, win_name, wait_time)

easycv/datasets/classification/raw.py

@@ -3,6 +3,7 @@
 import torch
 from PIL import Image
 
+from easycv.core.visualization.image import imshow_label
 from easycv.datasets.registry import DATASETS
 from easycv.datasets.shared.base import BaseDataset
 
@@ -59,3 +60,37 @@ def evaluate(self, results, evaluators, logger=None, topk=(1, 5)):
         eval_res = evaluators[0].evaluate(results, gt_labels)
 
         return eval_res
+
+    def visualize(self, results, vis_num=10, **kwargs):
+        """Visulaize the model output on validation data.
+        Args:
+            results: A dictionary containing
+                class: List of length number of test images.
+                img_metas: List of length number of test images,
+                    dict of image meta info, containing filename, img_shape,
+                    origin_img_shape and so on.
+            vis_num: number of images visualized
+        Returns: A dictionary containing
+            images: Visulaized images, list of np.ndarray.
+            img_metas: List of length number of test images,
+                    dict of image meta info, containing filename, img_shape,
+                    origin_img_shape and so on.
+        """
+        vis_imgs = []
+
+        # TODO: support img_metas for torch.jit
+        if results.get('img_metas', None) is None:
+            return {}
+
+        img_metas = results['img_metas'][:vis_num]
+        labels = results['class']
+
+        for i, img_meta in enumerate(img_metas):
+            filename = img_meta['filename']
+
+            vis_img = imshow_label(img=filename, labels=labels, show=False)
+            vis_imgs.append(vis_img)
+
+        output = {'images': vis_imgs, 'img_metas': img_metas}
+
+        return output

easycv/datasets/detection/mix.py

@@ -9,14 +9,13 @@
 import torch
 
 from easycv.datasets.registry import DATASETS, PIPELINES
-from easycv.datasets.shared.base import BaseDataset
 from easycv.utils import build_from_cfg
-from easycv.utils.bbox_util import batched_xyxy2cxcywh_with_shape
 from easycv.utils.bbox_util import xyxy2xywh as xyxy2cxcywh
+from .raw import DetDataset
 
 
 @DATASETS.register_module
-class DetImagesMixDataset(BaseDataset):
+class DetImagesMixDataset(DetDataset):
     """A wrapper of multiple images mixed dataset.
 
     Suitable for training on multiple images mixed data augmentation like
@@ -50,7 +49,7 @@ def __init__(self,
                  label_padding=True):
 
         super(DetImagesMixDataset, self).__init__(
-            data_source, pipeline, profiling=profiling)
+            data_source, pipeline, profiling=profiling, classes=classes)
 
         if skip_type_keys is not None:
             assert all([
@@ -70,10 +69,9 @@ def __init__(self,
             else:
                 raise TypeError('pipeline must be a dict')
 
-        self.CLASSES = classes
         if hasattr(self.data_source, 'flag'):
             self.flag = self.data_source.flag
-        self.num_samples = self.data_source.get_length()
+
         if dynamic_scale is not None:
             assert isinstance(dynamic_scale, tuple)
 
@@ -83,9 +81,6 @@ def __init__(self,
         self.label_padding = label_padding
         self.max_labels_num = 120
 
-    def __len__(self):
-        return self.num_samples
-
     def __getitem__(self, idx):
         results = copy.deepcopy(self.data_source.get_sample(idx))
         for (transform, transform_type) in zip(self.pipeline_yolox,
@@ -116,21 +111,8 @@ def __getitem__(self, idx):
             if 'img_scale' in results:
                 results.pop('img_scale')
 
-        # print(result.keys())
-
-        # if self.yolo_format:
-        #     # print(type(results['img_metas']), results['img_metas'])
-        #     # print(type(results['img_metas']._data), results['img_metas']._data)
-        #     img_shape = results['img_metas']._data['img_shape'][:2]
-        #     # print(type(results['gt_bboxes']))
-        #     gt_bboxes = xyxy2cxcywh_with_shape(results['gt_bboxes']._data, img_shape)
-        #     results['gt_bboxes'] = gt_bboxes.float()
-
         if self.label_padding:
-
             cxcywh_gt_bboxes = xyxy2cxcywh(results['gt_bboxes']._data)
-            # cxcywh_gt_bboxes = results['gt_bboxes']._data
-
             padded_gt_bboxes = torch.zeros((self.max_labels_num, 4),
                                            device=cxcywh_gt_bboxes.device)
             padded_gt_bboxes[range(cxcywh_gt_bboxes.shape[0])[:self.max_labels_num]] = \
@@ -146,9 +128,6 @@ def __getitem__(self, idx):
             results['gt_bboxes'] = padded_gt_bboxes
             results['gt_labels'] = padded_labels
 
-        # ['img_metas', 'img', 'gt_bboxes', 'gt_labels']
-        # results.pop('img_metas')
-        # print(results['img_metas'], "hhh", idx)
         return results
 
     def update_skip_type_keys(self, skip_type_keys):
@@ -240,30 +219,3 @@ def format_results(self, results, jsonfile_prefix=None, **kwargs):
             tmp_dir = None
         result_files = self.results2json(results, jsonfile_prefix)
         return result_files, tmp_dir
-
-    def evaluate(self, results, evaluators=None, logger=None):
-        '''results: a dict of list of Tensors, list length equals to number of test images
-        '''
-
-        eval_result = dict()
-
-        groundtruth_dict = {}
-        groundtruth_dict['groundtruth_boxes'] = [
-            batched_xyxy2cxcywh_with_shape(
-                self.data_source.get_ann_info(idx)['bboxes'],
-                results['img_metas'][idx]['ori_img_shape'])
-            for idx in range(len(results['img_metas']))
-        ]
-        groundtruth_dict['groundtruth_classes'] = [
-            self.data_source.get_ann_info(idx)['labels']
-            for idx in range(len(results['img_metas']))
-        ]
-        groundtruth_dict['groundtruth_is_crowd'] = [
-            self.data_source.get_ann_info(idx)['groundtruth_is_crowd']
-            for idx in range(len(results['img_metas']))
-        ]
-
-        for evaluator in evaluators:
-            eval_result.update(evaluator.evaluate(results, groundtruth_dict))
-        # print(eval_result)
-        return eval_result