PyTorch 中的 CocoDetection (1)

from torchvision.datasets import CocoDetectioncap_train2014_data = CocoDetection(    root="data/coco/imgs/train2014",    annFile="data/coco/anns/trainval2014/captions_train2014.json")cap_train2014_data = CocoDetection(    root="data/coco/imgs/train2014",    annFile="data/coco/anns/trainval2014/captions_train2014.json",    transform=None,    target_transform=None,    transforms=None)ins_train2014_data = CocoDetection(    root="data/coco/imgs/train2014",    annFile="data/coco/anns/trainval2014/instances_train2014.json")pk_train2014_data = CocoDetection(    root="data/coco/imgs/train2014",    annFile="data/coco/anns/trainval2014/person_keypoints_train2014.json")len(cap_train2014_data), len(ins_train2014_data), len(pk_train2014_data)# (82783, 82783, 82783)cap_val2014_data = CocoDetection(    root="data/coco/imgs/val2014",    annFile="data/coco/anns/trainval2014/captions_val2014.json")ins_val2014_data = CocoDetection(    root="data/coco/imgs/val2014",    annFile="data/coco/anns/trainval2014/instances_val2014.json")pk_val2014_data = CocoDetection(    root="data/coco/imgs/val2014",    annFile="data/coco/anns/trainval2014/person_keypoints_val2014.json")len(cap_val2014_data), len(ins_val2014_data), len(pk_val2014_data)# (40504, 40504, 40504)test2014_data = CocoDetection(    root="data/coco/imgs/test2014",    annFile="data/coco/anns/test2014/image_info_test2014.json")test2015_data = CocoDetection(    root="data/coco/imgs/test2015",    annFile="data/coco/anns/test2015/image_info_test2015.json")testdev2015_data = CocoDetection(    root="data/coco/imgs/test2015",    annFile="data/coco/anns/test2015/image_info_test-dev2015.json")len(test2014_data), len(test2015_data), len(testdev2015_data)# (40775, 81434, 20288)cap_train2014_data# Dataset CocoDetection#     Number of datapoints: 82783#     Root location: data/coco/imgs/train2014cap_train2014_data.root# 'data/coco/imgs/train2014'print(cap_train2014_data.transform)# Noneprint(cap_train2014_data.target_transform)# Noneprint(cap_train2014_data.transforms)# Nonecap_train2014_data.coco# <pycocotools.coco.COCO at 0x7c8a5f09d4f0>cap_train2014_data[26]# (<PIL.Image.Image image mode=RGB size=427x640>,#  [{'image_id': 154, 'id': 202466,#    'caption': 'three zeebras standing in a grassy field walking'},#   {'image_id': 154, 'id': 211904,#    'caption': 'Three zebras are standing in an open field.'},#   {'image_id': 154, 'id': 215654,#    'caption': 'Three zebra are walking through the grass of a field.'},#   {'image_id': 154, 'id': 216620,#    'caption': 'Three zebras standing on a grassy dirt field.'},#   {'image_id': 154, 'id': 231686,#    'caption': 'Three zebras grazing in green grass field area.'}])cap_train2014_data[179]# (<PIL.Image.Image image mode=RGB size=480x640>,#  [{'image_id': 1330, 'id': 721877,#    'caption': 'a young guy walking in a forrest holding ... his hand'},#   {'image_id': 1330, 'id': 727442,#    'caption': 'A partially black and white photo of a ... the woods.'},#   {'image_id': 1330, 'id': 730133,#    'caption': 'A disc golfer releases a throw ... wooded course.'},#   {'image_id': 1330, 'id': 731450,#    'caption': 'The person is in the clearing of a wooded area. '},#   {'image_id': 1330, 'id': 732335,#    'caption': 'a person throwing a frisbee at many trees '}])cap_train2014_data[194]# (<PIL.Image.Image image mode=RGB size=428x640>,#  [{'image_id': 1407, 'id': 451510,#    'caption': 'A person on a court with a tennis racket.'},#   {'image_id': 1407, 'id': 457735,#    'caption': 'A man that is holding a racquet ... the grass.'},#   {'image_id': 1407, 'id': 460600,#    'caption': 'A tennis player hits the ball during a match.'},#   {'image_id': 1407, 'id': 460612,#    'caption': 'The tennis player is poised to serve a ball.'},#   {'image_id': 1407, 'id': 821947,#    'caption': 'Man in white playing tennis on a court.'}])ins_train2014_data[26]# (<PIL.Image.Image image mode=RGB size=427x640>,#  [{'segmentation': [[229.5, 618.18, 235.64, ..., 219.85, 618.18]],#    'area': 53702.50415, 'iscrowd': 0, 'image_id': 154,#    'bbox': [11.98, 315.59, 349.08, 324.41], 'category_id': 24,#    'id': 590410},#   {'segmentation': ..., 'category_id': 24, 'id': 590623},#   {'segmentation': ..., 'category_id': 24, 'id': 593205}])ins_train2014_data[179]# (<PIL.Image.Image image mode=RGB size=480x640>,#  [{'segmentation': [[160.87, 574.0, 174.15, ..., 162.77, 577.6]],#    'area': 21922.32225, 'iscrowd': 0, 'image_id': 1330,#    'bbox': [38.47, 228.02, 249.55, 349.58], 'category_id': 1,#    'id': 497247},#   {'segmentation': ..., 'category_id': 34, 'id': 604179}])ins_train2014_data[194]# (<PIL.Image.Image image mode=RGB size=428x640>,#  [{'segmentation': [[203.26, 465.95, 215.13, ..., 207.22, 466.94]], #    'area': 20449.62315, 'iscrowd': 0, 'image_id': 1407,#    'bbox': [138.97, 198.88, 175.08, 355.11], 'category_id': 1,#    'id': 434962},#   {'segmentation': ..., 'category_id': 43, 'id': 658155},#   ...#   {'segmentation': ..., 'category_id': 1, 'id': 2000535}])pk_train2014_data[26]# (<PIL.Image.Image image mode=RGB size=427x640>, [])pk_train2014_data[179]# (<PIL.Image.Image image mode=RGB size=480x640>,#  [{'segmentation': [[160.87, 574, 174.15, ..., 162.77, 577.6]],#    'num_keypoints': 14, 'area': 21922.32225, 'iscrowd': 0,#    'keypoints': [0, 0, 0, 0, ..., 510, 2], 'image_id': 1330,#    'bbox': [38.47, 228.02, 249.55, 349.58], 'category_id': 1,#    'id': 497247}])pk_train2014_data[194]# (<PIL.Image.Image image mode=RGB size=428x640>,#  [{'segmentation': [[203.26, 465.95, 215.13, ..., 207.22, 466.94]],#    'num_keypoints': 16, 'area': 20449.62315, 'iscrowd': 0,#    'keypoints': [243, 289, 2, 247, ..., 516, 2], 'image_id': 1407,#    'bbox': [138.97, 198.88, 175.08, 355.11], 'category_id': 1,#    'id': 434962},#   {'segmentation': ..., 'category_id': 1, 'id': 1246131},#   ...#   {'segmentation': ..., 'category_id': 1, 'id': 2000535}])cap_val2014_data[26]# (<PIL.Image.Image image mode=RGB size=640x360>,#  [{'image_id': 428, 'id': 281051,#    'caption': 'a close up of a child next to a cake with balloons'},#   {'image_id': 428, 'id': 283808,#    'caption': 'A baby sitting in front of a cake wearing a tie.'},#   {'image_id': 428, 'id': 284135,#    'caption': 'The young boy is dressed in a tie that ... his cake. '},#  {'image_id': 428, 'id': 284627,#   'caption': 'A child eating a birthday cake near some balloons.'},#  {'image_id': 428, 'id': 401924,#   'caption': 'A baby eating a cake with a tie ... the background.'}])cap_val2014_data[179]# (<PIL.Image.Image image mode=RGB size=500x302>,#  [{'image_id': 2299, 'id': 692974,#    'caption': 'Many small children are posing ... white photo. '},#   {'image_id': 2299, 'id': 693640,#    'caption': 'A vintage school picture of grade school aged children.'},#   {'image_id': 2299, 'id': 694699,#    'caption': 'A black and white photo of a group of kids.'},#   {'image_id': 2299, 'id': 697432,#    'caption': 'A group of children standing next to each other.'},#   {'image_id': 2299, 'id': 698791,#    'caption': 'A group of children standing and ... each other. '}])cap_val2014_data[194]# (<PIL.Image.Image image mode=RGB size=640x427>,#  [{'image_id': 2562, 'id': 267259,#    'caption': 'A man hitting a tennis ball with a racquet.'},#   {'image_id': 2562, 'id': 277075,#    'caption': 'champion tennis player swats at the ball ... to win'},#   {'image_id': 2562, 'id': 279091,#    'caption': 'A man is hitting his tennis ball with ... the court.'},#   {'image_id': 2562, 'id': 406135,#    'caption': 'a tennis player on a court with a racket'},#   {'image_id': 2562, 'id': 823086,#    'caption': 'A professional tennis player hits a ... fans watch.'}])ins_val2014_data[26]# (<PIL.Image.Image image mode=RGB size=640x360>,#  [{'segmentation': [[378.61, 210.2, 409.35, ..., 374.56, 217.48]], #    'area': 3573.3858000000005, 'iscrowd': 0, 'image_id': 428,#    'bbox': [374.56, 200.49, 94.65, 154.52], 'category_id': 32,#    'id': 293908},#   {'segmentation': ..., 'category_id': 1, 'id': 487626},#   {'segmentation': ..., 'category_id': 61, 'id': 1085469}])ins_val2014_data[179]# (<PIL.Image.Image image mode=RGB size=500x302>,#  [{'segmentation': [[107.49, 226.51, 108.17, ..., 105.8, 226.43]],#    'area': 66.15510000000003, 'iscrowd': 0, 'image_id': 2299,#    'bbox': [101.74, 226.43, 7.53, 15.83], 'category_id': 32,#    'id': 295960},#   {'segmentation': ..., 'category_id': 32, 'id': 298359},#   ...#   {'segmentation': {'counts': [152, 13, 263, 40, 2, ..., 132, 75],#    'size': [302, 500]}, 'area': 87090, 'iscrowd': 1, 'image_id': 2299,#    'bbox': [0, 18, 499, 263], 'category_id': 1, 'id': 900100002299}])ins_val2014_data[194]# (<PIL.Image.Image image mode=RGB size=640x427>,#  [{'segmentation': [[389.92, 6.17, 391.48, ..., 393.57, 0.57]],#    'area': 482.5815999999996, 'iscrowd': 0, 'image_id': 2562,#    'bbox': [389.92, 0.57, 28.15, 21.38], 'category_id': 37,#    'id': 302161},#   {'segmentation': ..., 'category_id': 43, 'id': 659770},#   ...#   {'segmentation': {'counts': [132, 8, 370, 37, 3, ..., 82, 268],#    'size': [427, 640]}, 'area': 19849, 'iscrowd': 1, 'image_id': 2562, #    'bbox': [0, 49, 639, 193], 'category_id': 1, 'id': 900100002562}])pk_val2014_data[26]# (<PIL.Image.Image image mode=RGB size=640x360>,#  [{'segmentation': [[239.18, 244.08, 229.39, ..., 256.33, 251.43]],#    'num_keypoints': 10, 'area': 55007.0814, 'iscrowd': 0,#    'keypoints': [383, 132, 2, 418, ..., 0, 0], 'image_id': 428,#    'bbox': [226.94, 32.65, 355.92, 323.27], 'category_id': 1,#    'id': 487626}])pk_val2014_data[179]# (<PIL.Image.Image image mode=RGB size=500x302>,#  [{'segmentation': [[75, 272.02, 76.92, ..., 74.67, 272.66]],#    'num_keypoints': 17, 'area': 4357.5248, 'iscrowd': 0,#    'keypoints': [108, 213, 2, 113, ..., 289, 2], 'image_id': 2299,#    'bbox': [70.18, 189.51, 64.2, 112.04], 'category_id': 1,#    'id': 1219726},#   {'segmentation': ..., 'category_id': 1, 'id': 1226789},#   ...#   {'segmentation': {'counts': [152, 13, 263, 40, 2, ..., 132, 75],#    'size': [302, 500]}, 'num_keypoints': 0, 'area': 87090,#    'iscrowd': 1, 'keypoints': [0, 0, 0, 0, ..., 0, 0], 'image_id': 2299,#    'bbox': [0, 18, 499, 263], 'category_id': 1, 'id': 900100002299}])pk_val2014_data[194]# (<PIL.Image.Image image mode=RGB size=640x427>,#  [{'segmentation': [[19.26, 270.62, 4.3, ..., 25.98, 273.61]],#    'num_keypoints': 13, 'area': 6008.95835, 'iscrowd': 0,#    'keypoints': [60, 160, 2, 64, ..., 257, 1], 'image_id': 2562,#    'bbox': [4.3, 144.26, 100.19, 129.35], 'category_id': 1,#    'id': 1287168},#   {'segmentation': ..., 'category_id': 1, 'id': 1294190},#   ...#   {'segmentation': {'counts': [132, 8, 370, 37, 3, ..., 82, 268],#    'size': [427, 640]}, 'num_keypoints': 0, 'area': 19849, 'iscrowd': 1,#    'keypoints': [0, 0, 0, 0, ..., 0, 0], 'image_id': 2562,#    'bbox': [0, 49, 639, 193], 'category_id': 1, 'id': 900100002562}])test2014_data[26]# (<PIL.Image.Image image mode=RGB size=640x640>, [])test2014_data[179]# (<PIL.Image.Image image mode=RGB size=640x480>, [])test2014_data[194]# (<PIL.Image.Image image mode=RGB size=640x360>, [])test2015_data[26]# (<PIL.Image.Image image mode=RGB size=640x480>, [])test2015_data[179]# (<PIL.Image.Image image mode=RGB size=640x426>, [])test2015_data[194]# (<PIL.Image.Image image mode=RGB size=640x480>, [])testdev2015_data[26]# (<PIL.Image.Image image mode=RGB size=640x360>, [])testdev2015_data[179]# (<PIL.Image.Image image mode=RGB size=640x480>, [])testdev2015_data[194]# (<PIL.Image.Image image mode=RGB size=640x480>, [])import matplotlib.pyplot as pltfrom matplotlib.patches import Polygon, Rectangleimport numpy as npfrom pycocotools import mask# `show_images1()` doesn't work very well for the images with# segmentations and keypoints so for them, use `show_images2()` which# more uses the original coco functions. def show_images1(data, ims, main_title=None):    file = data.root.split('/')[-1]    fig, axes = plt.subplots(nrows=1, ncols=3, figsize=(14, 8))    fig.suptitle(t=main_title, y=0.9, fontsize=14)    x_crd = 0.02    for i, axis in zip(ims, axes.ravel()):        if data[i][1] and "caption" in data[i][1][0]:            im, anns = data[i]            axis.imshow(X=im)            axis.set_title(label=anns[0]["image_id"])            y_crd = 0.0            for ann in anns:                text_list = ann["caption"].split()                if len(text_list) > 9:                    text = " ".join(text_list[0:10]) + " ..."                else:                    text = " ".join(text_list)                plt.figtext(x=x_crd, y=y_crd, fontsize=10,                            s=f'{ann["id"]}:{text}')                y_crd -= 0.06            x_crd += 0.325            if i == 2 and file == "val2017":                x_crd += 0.06        if data[i][1] and "segmentation" in data[i][1][0]:            im, anns = data[i]            axis.imshow(X=im)            axis.set_title(label=anns[0]["image_id"])            for ann in anns:                if "counts" in ann['segmentation']:                    seg = ann['segmentation']                    # rle is Run Length Encoding.                    uncompressed_rle = [seg['counts']]                    height, width = seg['size']                    compressed_rle = mask.frPyObjects(pyobj=uncompressed_rle,                                                      h=height, w=width)                    # rld is Run Length Decoding.                    compressed_rld = mask.decode(rleObjs=compressed_rle)                    y_plts, x_plts = np.nonzero(a=np.squeeze(a=compressed_rld))                    axis.plot(x_plts, y_plts, color='yellow')                else:                    for seg in ann['segmentation']:                        seg_arrs = np.split(ary=np.array(seg),                                            indices_or_sections=len(seg)/2)                        poly = Polygon(xy=seg_arrs,                                       facecolor="lightgreen", alpha=0.7)                        axis.add_patch(p=poly)                        x_plts = [seg_arr[0] for seg_arr in seg_arrs]                        y_plts = [seg_arr[1] for seg_arr in seg_arrs]                        axis.plot(x_plts, y_plts, color='yellow')                x, y, w, h = ann['bbox']                rect = Rectangle(xy=(x, y), width=w, height=h,                                 linewidth=3, edgecolor='r',                                 facecolor='none', zorder=2)                axis.add_patch(p=rect)                if data[i][1] and 'keypoints' in data[i][1][0]:                    kps = ann['keypoints']                    kps_arrs = np.split(ary=np.array(kps),                                        indices_or_sections=len(kps)/3)                    x_plts = [kps_arr[0] for kps_arr in kps_arrs]                    y_plts = [kps_arr[1] for kps_arr in kps_arrs]                    nonzeros_x_plts = []                    nonzeros_y_plts = []                    for x_plt, y_plt in zip(x_plts, y_plts):                        if x_plt == 0 and y_plt == 0:                            continue                        nonzeros_x_plts.append(x_plt)                        nonzeros_y_plts.append(y_plt)                    axis.scatter(x=nonzeros_x_plts, y=nonzeros_y_plts,                                 color='yellow')                    # ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ Bad result ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓                    # axis.plot(nonzeros_x_plts, nonzeros_y_plts)        if not data[i][1]:            im, _ = data[i]            axis.imshow(X=im)    fig.tight_layout()    plt.show()ims = (26, 179, 194)show_images1(data=cap_train2014_data, ims=ims,             main_title="cap_train2014_data")show_images1(data=ins_train2014_data, ims=ims,             main_title="ins_train2014_data")show_images1(data=pk_train2014_data, ims=ims,             main_title="pk_train2014_data")print()show_images1(data=cap_val2014_data, ims=ims,              main_title="cap_val2014_data")show_images1(data=ins_val2014_data, ims=ims,              main_title="ins_val2014_data")show_images1(data=pk_val2014_data, ims=ims,             main_title="pk_val2014_data")print()show_images1(data=test2014_data, ims=ims,             main_title="test2014_data")show_images1(data=test2015_data, ims=ims,             main_title="test2015_data")show_images1(data=testdev2015_data, ims=ims,             main_title="testdev2015_data")# `show_images2()` works very well for the images with segmentations and# keypoints.def show_images2(data, index, main_title=None):    img_set = data[index]    img, img_anns = img_set    if img_anns and "segmentation" in img_anns[0]:        img_id = img_anns[0]['image_id']        coco = data.coco        def show_image(imgIds, areaRng=[],                       iscrowd=None, draw_bbox=False):            plt.figure(figsize=(11, 6))            plt.imshow(X=img)            plt.suptitle(t=main_title, y=1, fontsize=14)            plt.title(label=img_id, fontsize=14)            anns_ids = coco.getAnnIds(imgIds=img_id,                                      areaRng=areaRng, iscrowd=iscrowd)            anns = coco.loadAnns(ids=anns_ids)            coco.showAnns(anns=anns, draw_bbox=draw_bbox)            plt.show()        show_image(imgIds=img_id, draw_bbox=True)        show_image(imgIds=img_id, draw_bbox=False)        show_image(imgIds=img_id, iscrowd=False, draw_bbox=True)        show_image(imgIds=img_id, areaRng=[0, 5000], draw_bbox=True)    elif img_anns and not "segmentation" in img_anns[0]:        plt.figure(figsize=(11, 6))        img_id = img_anns[0]['image_id']        plt.imshow(X=img)        plt.suptitle(t=main_title, y=1, fontsize=14)        plt.title(label=img_id, fontsize=14)        plt.show()    elif not img_anns:        plt.figure(figsize=(11, 6))        plt.imshow(X=img)        plt.suptitle(t=main_title, y=1, fontsize=14)        plt.show()show_images2(data=ins_val2014_data, index=179,             main_title="ins_val2014_data")print()show_images2(data=pk_val2014_data, index=179,             main_title="pk_val2014_data")print()show_images2(data=ins_val2014_data, index=194,             main_title="ins_val2014_data")print()show_images2(data=pk_val2014_data, index=194,             main_title="pk_val2014_data")