Classification of one-inch maps

This is an example of a geospatial use of MapReader.

Download

https://mapreader.readthedocs.io/en/latest/User-guide/Download.html

[1]:

from mapreader import SheetDownloader

my_ts = SheetDownloader(
    metadata_path="../NLS_metadata/metadata_OS_One_Inch_GB_WFS_light.json",
    download_url="https://geo.nls.uk/maps/os/1inch_2nd_ed/{z}/{x}/{y}.png",
)

[INFO] Metadata file has 480 item(s).

[ ]:

my_ts.extract_wfs_id_nos()
my_ts.plot_all_metadata_on_map(map_extent="uk", add_id=False)

[3]:

my_ts.get_minmax_latlon()

[INFO] Min lat: 49.82091681, max lat: 60.97000003
[INFO] Min lon: -8.11999993, max lon: 1.83188116

[4]:

my_ts.hist_published_dates()

1873 1909 36

../_images/Worked-examples_one_inch_pipeline_5_1.png

Query maps using a list of lats/lons

[5]:

my_ts.query_map_sheets_by_coordinates((-4.33, 55.90))
my_ts.query_map_sheets_by_coordinates((-3.25, 51.93), append=True)

[6]:

my_ts.print_found_queries()

==============
Query results:
==============
URL:            https://maps.nls.uk/view/74488689
coordinates (bounds):   (-4.82999994, 55.79000003, -4.20999993, 56.05000003)
--------------------
URL:            https://maps.nls.uk/view/101168342
coordinates (bounds):   (-3.33544762, 51.87891401, -2.91228731, 52.05404182)
--------------------

[7]:

my_ts.plot_queries_on_map(map_extent="uk")

../_images/Worked-examples_one_inch_pipeline_9_0.png

Download map tiles

[8]:

my_ts.get_grid_bb(zoom_level=14)
my_ts.download_map_sheets_by_queries(overwrite=True)

[INFO] Downloaded "map_74488689.png"
[INFO] Downloaded "map_101168342.png"

Load

https://mapreader.readthedocs.io/en/latest/User-guide/Load.html

Load in downloaded maps

[9]:

from mapreader import loader

[10]:

my_files = loader("./maps/*png")

100%|█████████████████████████████████████████████████████████████████████████████| 2/2 [00:01<00:00,  1.14it/s]

[11]:

# len() shows the total number of images currently read (or sliced, see below)
print(f"Number of images: {len(my_files)}")

Number of images: 2

[12]:

print(my_files)

#images: 2

#parents: 2
map_74488689.png
map_101168342.png

#patches: 0

[13]:

my_files.add_metadata("./maps/metadata.csv")

[WARNING] Duplicates found in metadata. Keeping only first instance of each duplicated value

[15]:

parent_list = my_files.list_parents()

[16]:

my_files.show(parent_list[1], figsize=(20, 20))

100%|█████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00,  1.75it/s]

[16]:

[<Figure size 2000x2000 with 1 Axes>]

../_images/Worked-examples_one_inch_pipeline_20_2.png

Patchify maps

[17]:

my_files.patchify_all(patch_size=100, square_cuts=True)  # in pixels

[INFO] Saving patches in directory named "patches_100_pixel".

100%|█████████████████████████████████████████████████████████████████████████████| 2/2 [00:03<00:00,  1.89s/it]

[18]:

my_files.show_sample(num_samples=6, tree_level="patch")

../_images/Worked-examples_one_inch_pipeline_23_0.png

[19]:

my_files.show_parent(
    parent_list[0], patch_border=True, plot_parent=True, figsize=(15, 15)
)

100%|█████████████████████████████████████████████████████████████████████████████| 1/1 [00:02<00:00,  2.45s/it]

[19]:

[<Figure size 1500x1500 with 1 Axes>]

../_images/Worked-examples_one_inch_pipeline_24_2.png

Calculate pixel intensities

[20]:

my_files.calc_pixel_stats()

100%|█████████████████████████████████████████████████████████████████████████████| 2/2 [00:05<00:00,  2.65s/it]

[21]:

parent_df, patch_df = my_files.convert_images(save=True)

[INFO] Saved parent dataframe as "parent_df.csv"
[INFO] Saved patch dataframe as "patch_df.csv"

[22]:

parent_df.head()

[22]:

	parent_id	image_path	shape	name	url	coordinates	crs	published_date	grid_bb	patches	dlon	dlat
image_id
map_74488689.png	None	/Users/rwood/LwM/MapReader/worked_examples/geo...	(5632, 7424, 4)	map_74488689.png	https://maps.nls.uk/view/74488689	(-4.833984375, 55.80128097118045, -4.21875, 56...	EPSG:4326	1898	[(14, 7972, 5097)x(14, 8000, 5118)]	[patch-0-0-100-100-#map_74488689.png#.png, pat...	0.000083	0.000046
map_101168342.png	None	/Users/rwood/LwM/MapReader/worked_examples/geo...	(3584, 5120, 4)	map_101168342.png	https://maps.nls.uk/view/101168342	(-3.33984375, 51.89005393521691, -2.9223632812...	EPSG:4326	1896	[(14, 8040, 5407)x(14, 8059, 5420)]	[patch-0-0-100-100-#map_101168342.png#.png, pa...	0.000082	0.000049

[23]:

patch_df.head()

[23]:

	parent_id	image_path	shape	pixel_bounds	coordinates	crs	polygon	mean_pixel_R	mean_pixel_G	mean_pixel_B	mean_pixel_A	std_pixel_R	std_pixel_G	std_pixel_B	std_pixel_A
image_id
patch-0-0-100-100-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(100, 100, 4)	(0, 0, 100, 100)	(-4.833984375, 55.80128097118045, -4.825697274...	EPSG:4326	POLYGON ((-4.825697274043642 55.80128097118045...	0.882480	0.875822	0.814781	1.000000	0.092767	0.095471	0.091923	0.000000
patch-0-100-100-200-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(100, 100, 4)	(0, 100, 100, 200)	(-4.833984375, 55.80587061088084, -4.825697274...	EPSG:4326	POLYGON ((-4.825697274043642 55.80587061088084...	0.844518	0.836567	0.775709	0.999144	0.122864	0.126680	0.121811	0.020511
patch-0-200-100-300-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(100, 100, 4)	(0, 200, 100, 300)	(-4.833984375, 55.81046025058123, -4.825697274...	EPSG:4326	POLYGON ((-4.825697274043642 55.81046025058123...	0.795914	0.784561	0.719604	1.000000	0.104913	0.107458	0.103096	0.000000
patch-0-300-100-400-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(100, 100, 4)	(0, 300, 100, 400)	(-4.833984375, 55.81504989028163, -4.825697274...	EPSG:4326	POLYGON ((-4.825697274043642 55.81504989028163...	0.736126	0.723507	0.660471	1.000000	0.132483	0.136058	0.130438	0.000000
patch-0-400-100-500-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(100, 100, 4)	(0, 400, 100, 500)	(-4.833984375, 55.81963952998202, -4.825697274...	EPSG:4326	POLYGON ((-4.825697274043642 55.81963952998202...	0.783474	0.770846	0.705413	1.000000	0.127172	0.129634	0.123469	0.000000

[24]:

patch_df.mean()

[24]:

mean_pixel_R    0.795482
mean_pixel_G    0.787168
mean_pixel_B    0.725940
mean_pixel_A    0.999960
std_pixel_R     0.148695
std_pixel_G     0.151217
std_pixel_B     0.142944
std_pixel_A     0.000835
dtype: float64

[25]:

my_files.show_parent(
    parent_list[0],
    column_to_plot="mean_pixel_R",
    patch_border=False,
    plot_parent=True,
    vmin=0.75,
    vmax=0.9,
    figsize=(20, 20),
    alpha=0.5,
)

100%|█████████████████████████████████████████████████████████████████████████████| 1/1 [00:01<00:00,  1.48s/it]

[25]:

[<Figure size 2000x2000 with 2 Axes>]

../_images/Worked-examples_one_inch_pipeline_31_2.png

Annotate

https://mapreader.readthedocs.io/en/latest/User-guide/Annotate.html

Remember to first edit ./annotation_tasks.yaml file! It should look something like this:

# ---------------------------------------
# Define an annotation task
# This includes:
# 1. a name (e.g., building_simple or rail_space, see below)
# 2. a list of labels to be used for this task
# ---------------------------------------
tasks:
  building_simple:
    labels: ["No", "building"]
  rail_space:
    labels: ["No", "rail space"]

# ---------------------------------------
# paths
# You need to specify:
# 1. a name (e.g., task_test_one_inch_maps_001, see below)
# 2. patch_paths: path to all the patches to be annotated
# 3. parent_paths: path to the original/parent maps/images (which were patchified)
# 4. annot_dir: directory in which the outputs will be stored
# ---------------------------------------
paths:
  task_test_one_inch_maps_001:
    patch_paths: "./patches_100_pixel/patch-*png"
    parent_paths: "./maps/*png"
    annot_dir: "./annotations_one_inch"

Set up inputs

[1]:

from mapreader.annotate.utils import prepare_annotation, save_annotation

[5]:

userID = "rw"
annotation_tasks_file = "./annotation_tasks.yaml"
task = "rail_space"
annotation_set = "task_test_one_inch_maps_001"

# sortby="mean" sorts the patches according to the mean pixel intensities
# xoffset and yoffset specify the border size around a patch to be used as the context image (in pixel)
annotate = prepare_annotation(
    userID,
    task,
    annotation_tasks_file=annotation_tasks_file,
    annotation_set=annotation_set,
    sortby="mean",
    context_image=False,
)

0it [00:00, ?it/s]
100%|███████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 2937.19it/s]
100%|████████████████████████████████████████████████████████████████████| 6147/6147 [00:00<00:00, 18829.61it/s]
100%|█████████████████████████████████████████████████████████████████████████████| 2/2 [00:03<00:00,  1.53s/it]

Number of already annotated images: 186
Number of images to be annotated (total): 5961
Number of images to annotate (current batch): 100

Annotate maps and save annotations

[6]:

annotate

[7]:

save_annotation(
    annotate,
    userID,
    task,
    annotation_tasks_file=annotation_tasks_file,
    annotation_set=annotation_set,
)

[INFO] Save 53 new annotations to ./annotations_one_inch/rail_space_#rw#.csv
[INFO] 53 labels were not already stored
[INFO] Total number of saved annotations: 239

Train

https://mapreader.readthedocs.io/en/latest/User-guide/Classify.html

Read annotations

[35]:

from mapreader import AnnotationsLoader

[36]:

annotated_images = AnnotationsLoader()

annotated_images.load("./annotations_one_inch/rail_space_#rw#.csv")

[INFO] Reading "./annotations_one_inch/rail_space_#rw#.csv"
[INFO] Number of annotations:   239

[INFO] Number of instances of each label (from column "label"):
        - 1:      182
        - 2:      57

[37]:

annotated_images.annotations

[37]:

	image_id	image_path	label	label_index
0	patch-2900-1900-3000-2000-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1	0
1	patch-6900-4300-7000-4400-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1	0
2	patch-6600-4200-6700-4300-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1	0
3	patch-6600-4100-6700-4200-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1	0
4	patch-0-2100-100-2200-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1	0
...	...	...	...	...
234	patch-3400-5300-3500-5400-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1	0
235	patch-2600-2000-2700-2100-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1	0
236	patch-4500-3484-4600-3584-#map_101168342.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1	0
237	patch-4500-2100-4600-2200-#map_101168342.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1	0
238	patch-2200-2600-2300-2700-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	2	1

239 rows × 4 columns

[11]:

print(annotated_images)

[INFO] Number of annotations:   239

[INFO] Number of instances of each label (from column "label"):
        - 1:      182
        - 2:      57

[12]:

# show sample images for target label (tar_label)
annotated_images.show_sample(label_to_show="2")

../_images/Worked-examples_one_inch_pipeline_46_0.png

[13]:

# show an image based on its index
annotated_images.show_patch("patch-3800-3200-3900-3300-#map_101168342.png#.png")

../_images/Worked-examples_one_inch_pipeline_47_0.png

[14]:

annotated_images.review_labels()

[INFO] Type "exit", "end" or "stop" to exit.
[INFO] Showing 0-24 out of 239.

Enter IDs, comma separated (or press enter to continue): exit
[INFO] Exited.

Prepare datasets and dataloaders

[38]:

annotated_images.create_datasets(frac_train=0.7, frac_val=0.2, frac_test=0.1)

[INFO] Number of annotations in each set:
        - Train:        167
        - Validate:     48
        - Test:         24

[39]:

annotated_images.datasets["train"].patch_df.head()

[39]:

	image_id	image_path	label
184	patch-4700-300-4800-400-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1
78	patch-1900-2500-2000-2600-#map_101168342.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1
191	patch-3500-3100-3600-3200-#map_101168342.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1
180	patch-1500-1100-1600-1200-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1
36	patch-6200-3100-6300-3200-#map_74488689.png#.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	1

[40]:

annotated_images.labels_map

[40]:

{0: '1', 1: '2'}

[41]:

dataloaders = annotated_images.create_dataloaders(batch_size=8)

[INFO] Using default sampler.

Set up `ClassifierContainer`

Load a pretrained PyTorch model, dataloaders and labels map

[42]:

from mapreader import ClassifierContainer

[43]:

my_classifier = ClassifierContainer(model ="resnet18",
                                    labels_map={0: 'No', 1: 'rail space'},
                                    dataloaders=dataloaders
                                    )

[INFO] Device is set to cpu
[INFO] Loaded "train" with 167 items.
[INFO] Loaded "val" with 48 items.
[INFO] Loaded "test" with 24 items.
[INFO] Initializing model.

[44]:

my_classifier.labels_map

[44]:

{0: 'No', 1: 'rail space'}

[45]:

my_classifier.show_sample(set_name="train", batch_number=1)

[INFO] dataset: train
        - items:        167
        - batch size:   8
        - batches:      21

../_images/Worked-examples_one_inch_pipeline_59_1.png

[46]:

my_classifier.model_summary()

===================================================================================================================
Layer (type:depth-idx)                   Output Shape              Output Shape              Param #
===================================================================================================================
ResNet                                   [8, 2]                    [8, 2]                    --
├─Conv2d: 1-1                            [8, 64, 112, 112]         [8, 64, 112, 112]         9,408
├─BatchNorm2d: 1-2                       [8, 64, 112, 112]         [8, 64, 112, 112]         128
├─ReLU: 1-3                              [8, 64, 112, 112]         [8, 64, 112, 112]         --
├─MaxPool2d: 1-4                         [8, 64, 56, 56]           [8, 64, 56, 56]           --
├─Sequential: 1-5                        [8, 64, 56, 56]           [8, 64, 56, 56]           --
│    └─BasicBlock: 2-1                   [8, 64, 56, 56]           [8, 64, 56, 56]           --
│    │    └─Conv2d: 3-1                  [8, 64, 56, 56]           [8, 64, 56, 56]           36,864
│    │    └─BatchNorm2d: 3-2             [8, 64, 56, 56]           [8, 64, 56, 56]           128
│    │    └─ReLU: 3-3                    [8, 64, 56, 56]           [8, 64, 56, 56]           --
│    │    └─Conv2d: 3-4                  [8, 64, 56, 56]           [8, 64, 56, 56]           36,864
│    │    └─BatchNorm2d: 3-5             [8, 64, 56, 56]           [8, 64, 56, 56]           128
│    │    └─ReLU: 3-6                    [8, 64, 56, 56]           [8, 64, 56, 56]           --
│    └─BasicBlock: 2-2                   [8, 64, 56, 56]           [8, 64, 56, 56]           --
│    │    └─Conv2d: 3-7                  [8, 64, 56, 56]           [8, 64, 56, 56]           36,864
│    │    └─BatchNorm2d: 3-8             [8, 64, 56, 56]           [8, 64, 56, 56]           128
│    │    └─ReLU: 3-9                    [8, 64, 56, 56]           [8, 64, 56, 56]           --
│    │    └─Conv2d: 3-10                 [8, 64, 56, 56]           [8, 64, 56, 56]           36,864
│    │    └─BatchNorm2d: 3-11            [8, 64, 56, 56]           [8, 64, 56, 56]           128
│    │    └─ReLU: 3-12                   [8, 64, 56, 56]           [8, 64, 56, 56]           --
├─Sequential: 1-6                        [8, 128, 28, 28]          [8, 128, 28, 28]          --
│    └─BasicBlock: 2-3                   [8, 128, 28, 28]          [8, 128, 28, 28]          --
│    │    └─Conv2d: 3-13                 [8, 128, 28, 28]          [8, 128, 28, 28]          73,728
│    │    └─BatchNorm2d: 3-14            [8, 128, 28, 28]          [8, 128, 28, 28]          256
│    │    └─ReLU: 3-15                   [8, 128, 28, 28]          [8, 128, 28, 28]          --
│    │    └─Conv2d: 3-16                 [8, 128, 28, 28]          [8, 128, 28, 28]          147,456
│    │    └─BatchNorm2d: 3-17            [8, 128, 28, 28]          [8, 128, 28, 28]          256
│    │    └─Sequential: 3-18             [8, 128, 28, 28]          [8, 128, 28, 28]          8,448
│    │    └─ReLU: 3-19                   [8, 128, 28, 28]          [8, 128, 28, 28]          --
│    └─BasicBlock: 2-4                   [8, 128, 28, 28]          [8, 128, 28, 28]          --
│    │    └─Conv2d: 3-20                 [8, 128, 28, 28]          [8, 128, 28, 28]          147,456
│    │    └─BatchNorm2d: 3-21            [8, 128, 28, 28]          [8, 128, 28, 28]          256
│    │    └─ReLU: 3-22                   [8, 128, 28, 28]          [8, 128, 28, 28]          --
│    │    └─Conv2d: 3-23                 [8, 128, 28, 28]          [8, 128, 28, 28]          147,456
│    │    └─BatchNorm2d: 3-24            [8, 128, 28, 28]          [8, 128, 28, 28]          256
│    │    └─ReLU: 3-25                   [8, 128, 28, 28]          [8, 128, 28, 28]          --
├─Sequential: 1-7                        [8, 256, 14, 14]          [8, 256, 14, 14]          --
│    └─BasicBlock: 2-5                   [8, 256, 14, 14]          [8, 256, 14, 14]          --
│    │    └─Conv2d: 3-26                 [8, 256, 14, 14]          [8, 256, 14, 14]          294,912
│    │    └─BatchNorm2d: 3-27            [8, 256, 14, 14]          [8, 256, 14, 14]          512
│    │    └─ReLU: 3-28                   [8, 256, 14, 14]          [8, 256, 14, 14]          --
│    │    └─Conv2d: 3-29                 [8, 256, 14, 14]          [8, 256, 14, 14]          589,824
│    │    └─BatchNorm2d: 3-30            [8, 256, 14, 14]          [8, 256, 14, 14]          512
│    │    └─Sequential: 3-31             [8, 256, 14, 14]          [8, 256, 14, 14]          33,280
│    │    └─ReLU: 3-32                   [8, 256, 14, 14]          [8, 256, 14, 14]          --
│    └─BasicBlock: 2-6                   [8, 256, 14, 14]          [8, 256, 14, 14]          --
│    │    └─Conv2d: 3-33                 [8, 256, 14, 14]          [8, 256, 14, 14]          589,824
│    │    └─BatchNorm2d: 3-34            [8, 256, 14, 14]          [8, 256, 14, 14]          512
│    │    └─ReLU: 3-35                   [8, 256, 14, 14]          [8, 256, 14, 14]          --
│    │    └─Conv2d: 3-36                 [8, 256, 14, 14]          [8, 256, 14, 14]          589,824
│    │    └─BatchNorm2d: 3-37            [8, 256, 14, 14]          [8, 256, 14, 14]          512
│    │    └─ReLU: 3-38                   [8, 256, 14, 14]          [8, 256, 14, 14]          --
├─Sequential: 1-8                        [8, 512, 7, 7]            [8, 512, 7, 7]            --
│    └─BasicBlock: 2-7                   [8, 512, 7, 7]            [8, 512, 7, 7]            --
│    │    └─Conv2d: 3-39                 [8, 512, 7, 7]            [8, 512, 7, 7]            1,179,648
│    │    └─BatchNorm2d: 3-40            [8, 512, 7, 7]            [8, 512, 7, 7]            1,024
│    │    └─ReLU: 3-41                   [8, 512, 7, 7]            [8, 512, 7, 7]            --
│    │    └─Conv2d: 3-42                 [8, 512, 7, 7]            [8, 512, 7, 7]            2,359,296
│    │    └─BatchNorm2d: 3-43            [8, 512, 7, 7]            [8, 512, 7, 7]            1,024
│    │    └─Sequential: 3-44             [8, 512, 7, 7]            [8, 512, 7, 7]            132,096
│    │    └─ReLU: 3-45                   [8, 512, 7, 7]            [8, 512, 7, 7]            --
│    └─BasicBlock: 2-8                   [8, 512, 7, 7]            [8, 512, 7, 7]            --
│    │    └─Conv2d: 3-46                 [8, 512, 7, 7]            [8, 512, 7, 7]            2,359,296
│    │    └─BatchNorm2d: 3-47            [8, 512, 7, 7]            [8, 512, 7, 7]            1,024
│    │    └─ReLU: 3-48                   [8, 512, 7, 7]            [8, 512, 7, 7]            --
│    │    └─Conv2d: 3-49                 [8, 512, 7, 7]            [8, 512, 7, 7]            2,359,296
│    │    └─BatchNorm2d: 3-50            [8, 512, 7, 7]            [8, 512, 7, 7]            1,024
│    │    └─ReLU: 3-51                   [8, 512, 7, 7]            [8, 512, 7, 7]            --
├─AdaptiveAvgPool2d: 1-9                 [8, 512, 1, 1]            [8, 512, 1, 1]            --
├─Linear: 1-10                           [8, 2]                    [8, 2]                    1,026
===================================================================================================================
Total params: 11,177,538
Trainable params: 11,177,538
Non-trainable params: 0
Total mult-adds (G): 14.51
===================================================================================================================
Input size (MB): 4.82
Forward/backward pass size (MB): 317.92
Params size (MB): 44.71
Estimated Total Size (MB): 367.44
===================================================================================================================

Set up criterion, optimizer and scheduler

[47]:

my_classifier.add_criterion("cross-entropy")

[INFO] Using "CrossEntropyLoss()" as criterion.

[48]:

params_to_optimize = my_classifier.generate_layerwise_lrs(
    min_lr=1e-4, max_lr=1e-3, spacing="geomspace"
)

[49]:

my_classifier.initialize_optimizer(params2optimize=params_to_optimize)

[50]:

my_classifier.initialize_scheduler()

Train/fine-tune model

Note: it is possible to interrupt a training (using Kernel/Interrupt in Jupyter Notebook or ctrl+C).

[51]:

my_classifier.train(
    num_epochs=10,
    save_model_dir="./models_tutorial",
    tensorboard_path="tboard_tutorial",
    tmp_file_save_freq=2,
    remove_after_load=False,
    print_info_batch_freq=10,
)

[INFO] Each training step will pass: ['train', 'val'].
2023-06-05 13:28:03 599-JY5FK6 [INFO] train    -- 1/10 --      8/167 (  4.8% ) -- Loss: 0.607
2023-06-05 13:28:06 599-JY5FK6 [INFO] train    -- 1/10 --     88/167 ( 52.7% ) -- Loss: 3.242
2023-06-05 13:28:09 599-JY5FK6 [INFO] train    -- 1/10 --    167/167 (100.0% ) -- Loss: 0.467
2023-06-05 13:28:09 599-JY5FK6 [INFO] train    -- 1/10 -- Loss: 1.135; F_macro: 65.27; R_macro: 65.36
2023-06-05 13:28:09 599-JY5FK6 [INFO] val      -- 1/10 --       8/48 ( 16.7% ) -- Loss: 8.293
2023-06-05 13:28:09 599-JY5FK6 [INFO] val      -- 1/10 -- Loss: 5.942; F_macro: 21.60; R_macro: 51.35

2023-06-05 13:28:10 599-JY5FK6 [INFO] train    -- 2/10 --      8/167 (  4.8% ) -- Loss: 0.856
2023-06-05 13:28:12 599-JY5FK6 [INFO] train    -- 2/10 --     88/167 ( 52.7% ) -- Loss: 0.592
2023-06-05 13:28:15 599-JY5FK6 [INFO] train    -- 2/10 --    167/167 (100.0% ) -- Loss: 0.393
2023-06-05 13:28:15 599-JY5FK6 [INFO] train    -- 2/10 -- Loss: 0.553; F_macro: 74.84; R_macro: 75.45
2023-06-05 13:28:15 599-JY5FK6 [INFO] val      -- 2/10 --       8/48 ( 16.7% ) -- Loss: 0.958
2023-06-05 13:28:15 599-JY5FK6 [INFO] val      -- 2/10 -- Loss: 1.185; F_macro: 55.56; R_macro: 63.39

[INFO] Checkpoint file saved to "./tmp_checkpoints/tmp_8981877014_checkpoint.pkl".
2023-06-05 13:28:16 599-JY5FK6 [INFO] train    -- 3/10 --      8/167 (  4.8% ) -- Loss: 0.202
2023-06-05 13:28:18 599-JY5FK6 [INFO] train    -- 3/10 --     88/167 ( 52.7% ) -- Loss: 0.290
2023-06-05 13:28:21 599-JY5FK6 [INFO] train    -- 3/10 --    167/167 (100.0% ) -- Loss: 0.545
2023-06-05 13:28:21 599-JY5FK6 [INFO] train    -- 3/10 -- Loss: 0.375; F_macro: 84.81; R_macro: 84.76
2023-06-05 13:28:21 599-JY5FK6 [INFO] val      -- 3/10 --       8/48 ( 16.7% ) -- Loss: 0.197
2023-06-05 13:28:22 599-JY5FK6 [INFO] val      -- 3/10 -- Loss: 0.366; F_macro: 82.31; R_macro: 82.31

2023-06-05 13:28:22 599-JY5FK6 [INFO] train    -- 4/10 --      8/167 (  4.8% ) -- Loss: 1.635
2023-06-05 13:28:25 599-JY5FK6 [INFO] train    -- 4/10 --     88/167 ( 52.7% ) -- Loss: 0.111
2023-06-05 13:28:27 599-JY5FK6 [INFO] train    -- 4/10 --    167/167 (100.0% ) -- Loss: 0.103
2023-06-05 13:28:27 599-JY5FK6 [INFO] train    -- 4/10 -- Loss: 0.407; F_macro: 86.23; R_macro: 86.60
2023-06-05 13:28:27 599-JY5FK6 [INFO] val      -- 4/10 --       8/48 ( 16.7% ) -- Loss: 0.234
2023-06-05 13:28:28 599-JY5FK6 [INFO] val      -- 4/10 -- Loss: 0.392; F_macro: 79.49; R_macro: 75.92

[INFO] Checkpoint file saved to "./tmp_checkpoints/tmp_8981877014_checkpoint.pkl".
2023-06-05 13:28:28 599-JY5FK6 [INFO] train    -- 5/10 --      8/167 (  4.8% ) -- Loss: 0.169
2023-06-05 13:28:31 599-JY5FK6 [INFO] train    -- 5/10 --     88/167 ( 52.7% ) -- Loss: 0.181
2023-06-05 13:28:34 599-JY5FK6 [INFO] train    -- 5/10 --    167/167 (100.0% ) -- Loss: 0.683
2023-06-05 13:28:34 599-JY5FK6 [INFO] train    -- 5/10 -- Loss: 0.268; F_macro: 90.99; R_macro: 90.92
2023-06-05 13:28:34 599-JY5FK6 [INFO] val      -- 5/10 --       8/48 ( 16.7% ) -- Loss: 0.175
2023-06-05 13:28:34 599-JY5FK6 [INFO] val      -- 5/10 -- Loss: 0.747; F_macro: 67.83; R_macro: 65.48

2023-06-05 13:28:35 599-JY5FK6 [INFO] train    -- 6/10 --      8/167 (  4.8% ) -- Loss: 1.160
2023-06-05 13:28:37 599-JY5FK6 [INFO] train    -- 6/10 --     88/167 ( 52.7% ) -- Loss: 0.273
2023-06-05 13:28:40 599-JY5FK6 [INFO] train    -- 6/10 --    167/167 (100.0% ) -- Loss: 0.240
2023-06-05 13:28:40 599-JY5FK6 [INFO] train    -- 6/10 -- Loss: 0.397; F_macro: 85.60; R_macro: 85.65
2023-06-05 13:28:40 599-JY5FK6 [INFO] val      -- 6/10 --       8/48 ( 16.7% ) -- Loss: 0.693
2023-06-05 13:28:40 599-JY5FK6 [INFO] val      -- 6/10 -- Loss: 1.080; F_macro: 58.97; R_macro: 71.13

[INFO] Checkpoint file saved to "./tmp_checkpoints/tmp_8981877014_checkpoint.pkl".
2023-06-05 13:28:41 599-JY5FK6 [INFO] train    -- 7/10 --      8/167 (  4.8% ) -- Loss: 0.343
2023-06-05 13:28:43 599-JY5FK6 [INFO] train    -- 7/10 --     88/167 ( 52.7% ) -- Loss: 0.329
2023-06-05 13:28:46 599-JY5FK6 [INFO] train    -- 7/10 --    167/167 (100.0% ) -- Loss: 1.072
2023-06-05 13:28:46 599-JY5FK6 [INFO] train    -- 7/10 -- Loss: 0.438; F_macro: 82.27; R_macro: 81.94
2023-06-05 13:28:46 599-JY5FK6 [INFO] val      -- 7/10 --       8/48 ( 16.7% ) -- Loss: 1.808
2023-06-05 13:28:47 599-JY5FK6 [INFO] val      -- 7/10 -- Loss: 1.582; F_macro: 49.65; R_macro: 64.37

2023-06-05 13:28:47 599-JY5FK6 [INFO] train    -- 8/10 --      8/167 (  4.8% ) -- Loss: 0.362
2023-06-05 13:28:50 599-JY5FK6 [INFO] train    -- 8/10 --     88/167 ( 52.7% ) -- Loss: 1.773
2023-06-05 13:28:52 599-JY5FK6 [INFO] train    -- 8/10 --    167/167 (100.0% ) -- Loss: 0.381
2023-06-05 13:28:52 599-JY5FK6 [INFO] train    -- 8/10 -- Loss: 0.475; F_macro: 78.42; R_macro: 78.46
2023-06-05 13:28:52 599-JY5FK6 [INFO] val      -- 8/10 --       8/48 ( 16.7% ) -- Loss: 0.606
2023-06-05 13:28:53 599-JY5FK6 [INFO] val      -- 8/10 -- Loss: 0.879; F_macro: 72.83; R_macro: 78.75

[INFO] Checkpoint file saved to "./tmp_checkpoints/tmp_8981877014_checkpoint.pkl".
2023-06-05 13:28:53 599-JY5FK6 [INFO] train    -- 9/10 --      8/167 (  4.8% ) -- Loss: 0.431
2023-06-05 13:28:56 599-JY5FK6 [INFO] train    -- 9/10 --     88/167 ( 52.7% ) -- Loss: 1.026
2023-06-05 13:28:59 599-JY5FK6 [INFO] train    -- 9/10 --    167/167 (100.0% ) -- Loss: 0.376
2023-06-05 13:28:59 599-JY5FK6 [INFO] train    -- 9/10 -- Loss: 0.457; F_macro: 84.40; R_macro: 84.53
2023-06-05 13:28:59 599-JY5FK6 [INFO] val      -- 9/10 --       8/48 ( 16.7% ) -- Loss: 2.823
2023-06-05 13:28:59 599-JY5FK6 [INFO] val      -- 9/10 -- Loss: 1.926; F_macro: 34.71; R_macro: 58.11

2023-06-05 13:28:59 599-JY5FK6 [INFO] train    -- 10/10 --      8/167 (  4.8% ) -- Loss: 1.207
2023-06-05 13:29:02 599-JY5FK6 [INFO] train    -- 10/10 --     88/167 ( 52.7% ) -- Loss: 0.278
2023-06-05 13:29:05 599-JY5FK6 [INFO] train    -- 10/10 --    167/167 (100.0% ) -- Loss: 0.601
2023-06-05 13:29:05 599-JY5FK6 [INFO] train    -- 10/10 -- Loss: 0.465; F_macro: 78.67; R_macro: 78.63
2023-06-05 13:29:05 599-JY5FK6 [INFO] val      -- 10/10 --       8/48 ( 16.7% ) -- Loss: 0.385
2023-06-05 13:29:05 599-JY5FK6 [INFO] val      -- 10/10 -- Loss: 0.584; F_macro: 52.38; R_macro: 54.55

[INFO] Checkpoint file saved to "./tmp_checkpoints/tmp_8981877014_checkpoint.pkl".
[INFO] Total time: 1m 2s
[INFO] Model at epoch 3 has least valid loss (0.3659) so will be saved.
[INFO] Path: /Users/rwood/LwM/MapReader/worked_examples/geospatial/classification_one_inch_maps/models_tutorial/checkpoint_3.pkl

[52]:

list(my_classifier.metrics.keys())

[52]:

['epoch_loss_train',
 'epoch_prec_micro_train',
 'epoch_recall_micro_train',
 'epoch_fscore_micro_train',
 'epoch_supp_micro_train',
 'epoch_rocauc_micro_train',
 'epoch_prec_macro_train',
 'epoch_recall_macro_train',
 'epoch_fscore_macro_train',
 'epoch_supp_macro_train',
 'epoch_rocauc_macro_train',
 'epoch_prec_weighted_train',
 'epoch_recall_weighted_train',
 'epoch_fscore_weighted_train',
 'epoch_supp_weighted_train',
 'epoch_rocauc_weighted_train',
 'epoch_prec_0_train',
 'epoch_recall_0_train',
 'epoch_fscore_0_train',
 'epoch_supp_0_train',
 'epoch_prec_1_train',
 'epoch_recall_1_train',
 'epoch_fscore_1_train',
 'epoch_supp_1_train',
 'epoch_loss_val',
 'epoch_prec_micro_val',
 'epoch_recall_micro_val',
 'epoch_fscore_micro_val',
 'epoch_supp_micro_val',
 'epoch_rocauc_micro_val',
 'epoch_prec_macro_val',
 'epoch_recall_macro_val',
 'epoch_fscore_macro_val',
 'epoch_supp_macro_val',
 'epoch_rocauc_macro_val',
 'epoch_prec_weighted_val',
 'epoch_recall_weighted_val',
 'epoch_fscore_weighted_val',
 'epoch_supp_weighted_val',
 'epoch_rocauc_weighted_val',
 'epoch_prec_0_val',
 'epoch_recall_0_val',
 'epoch_fscore_0_val',
 'epoch_supp_0_val',
 'epoch_prec_1_val',
 'epoch_recall_1_val',
 'epoch_fscore_1_val',
 'epoch_supp_1_val']

[53]:

my_classifier.plot_metric(
    y_axis=["epoch_loss_train", "epoch_loss_val"],
    y_label="Loss",
    legends=["Train", "Valid"],
)

../_images/Worked-examples_one_inch_pipeline_70_0.png

[54]:

my_classifier.plot_metric(
    y_axis=["epoch_rocauc_macro_train", "epoch_rocauc_macro_val"],
    y_label="ROC AUC",
    legends=["Train", "Valid"],
    colors=["k", "tab:red"],
)

../_images/Worked-examples_one_inch_pipeline_71_0.png

[55]:

my_classifier.plot_metric(
    y_axis=[
        "epoch_recall_macro_train",
        "epoch_recall_macro_val",
        "epoch_recall_0_val",
        "epoch_recall_1_val",
    ],
    y_label="Recall",
    legends=[
        "Train",
        "Valid",
        "Valid (label: 0)",
        "Valid (label: 1)",
    ],
    colors=["k", "tab:red", "tab:red", "tab:red"],
    styles=["-", "-", "--", ":"],
    markers=["o", "o", "", ""],
    plt_yrange=[0, 100],
)

../_images/Worked-examples_one_inch_pipeline_72_0.png

Model inference

Here, we show how to load an already trained/fine-tuned model and a dataset and then do model inference.

Load patches and add metadata

[56]:

from mapreader import load_patches

[57]:

my_maps = load_patches(
    "./patches_100_pixel/*74488689*png", parent_paths="./maps/map_74488689.png"
)

my_maps.add_metadata("./maps/metadata.csv", ignore_mismatch=True)
my_maps.add_metadata("patch_df.csv", tree_level="patch", ignore_mismatch=True)

0it [00:00, ?it/s]
100%|███████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 2132.34it/s]
100%|████████████████████████████████████████████████████████████████████| 4275/4275 [00:00<00:00, 18332.24it/s]

[WARNING] Duplicates found in metadata. Keeping only first instance of each duplicated value

[58]:

parent_df, patch_df = my_maps.convert_images()

Create dataset from patches

[59]:

from mapreader import PatchDataset

[60]:

patch_dataset = PatchDataset(patch_df, transform="val")

[61]:

my_classifier.load_dataset(
    patch_dataset, set_name="patches", batch_size=8, shuffle=False
)

[62]:

my_classifier.print_batch_info(set_name="patches")

[INFO] dataset: patches
        - items:        4275
        - batch size:   8
        - batches:      535

Inference

[63]:

my_classifier.inference(set_name="patches")

[INFO] Each training step will pass: ['patches'].
2023-06-05 13:29:08 599-JY5FK6 [INFO] patches  -- 11/11 --     8/4275 (  0.2% ) -- 
2023-06-05 13:29:08 599-JY5FK6 [INFO] patches  -- 11/11 --    48/4275 (  1.1% ) -- 
2023-06-05 13:29:09 599-JY5FK6 [INFO] patches  -- 11/11 --    88/4275 (  2.1% ) -- 
2023-06-05 13:29:09 599-JY5FK6 [INFO] patches  -- 11/11 --   128/4275 (  3.0% ) -- 
2023-06-05 13:29:10 599-JY5FK6 [INFO] patches  -- 11/11 --   168/4275 (  3.9% ) -- 
2023-06-05 13:29:10 599-JY5FK6 [INFO] patches  -- 11/11 --   208/4275 (  4.9% ) -- 
2023-06-05 13:29:11 599-JY5FK6 [INFO] patches  -- 11/11 --   248/4275 (  5.8% ) -- 
2023-06-05 13:29:11 599-JY5FK6 [INFO] patches  -- 11/11 --   288/4275 (  6.7% ) -- 
2023-06-05 13:29:11 599-JY5FK6 [INFO] patches  -- 11/11 --   328/4275 (  7.7% ) -- 
2023-06-05 13:29:12 599-JY5FK6 [INFO] patches  -- 11/11 --   368/4275 (  8.6% ) -- 
2023-06-05 13:29:12 599-JY5FK6 [INFO] patches  -- 11/11 --   408/4275 (  9.5% ) -- 
2023-06-05 13:29:13 599-JY5FK6 [INFO] patches  -- 11/11 --   448/4275 ( 10.5% ) -- 
2023-06-05 13:29:13 599-JY5FK6 [INFO] patches  -- 11/11 --   488/4275 ( 11.4% ) -- 
2023-06-05 13:29:14 599-JY5FK6 [INFO] patches  -- 11/11 --   528/4275 ( 12.4% ) -- 
2023-06-05 13:29:14 599-JY5FK6 [INFO] patches  -- 11/11 --   568/4275 ( 13.3% ) -- 
2023-06-05 13:29:15 599-JY5FK6 [INFO] patches  -- 11/11 --   608/4275 ( 14.2% ) -- 
2023-06-05 13:29:15 599-JY5FK6 [INFO] patches  -- 11/11 --   648/4275 ( 15.2% ) -- 
2023-06-05 13:29:15 599-JY5FK6 [INFO] patches  -- 11/11 --   688/4275 ( 16.1% ) -- 
2023-06-05 13:29:16 599-JY5FK6 [INFO] patches  -- 11/11 --   728/4275 ( 17.0% ) -- 
2023-06-05 13:29:16 599-JY5FK6 [INFO] patches  -- 11/11 --   768/4275 ( 18.0% ) -- 
2023-06-05 13:29:17 599-JY5FK6 [INFO] patches  -- 11/11 --   808/4275 ( 18.9% ) -- 
2023-06-05 13:29:17 599-JY5FK6 [INFO] patches  -- 11/11 --   848/4275 ( 19.8% ) -- 
2023-06-05 13:29:18 599-JY5FK6 [INFO] patches  -- 11/11 --   888/4275 ( 20.8% ) -- 
2023-06-05 13:29:18 599-JY5FK6 [INFO] patches  -- 11/11 --   928/4275 ( 21.7% ) -- 
2023-06-05 13:29:18 599-JY5FK6 [INFO] patches  -- 11/11 --   968/4275 ( 22.6% ) -- 
2023-06-05 13:29:19 599-JY5FK6 [INFO] patches  -- 11/11 --  1008/4275 ( 23.6% ) -- 
2023-06-05 13:29:19 599-JY5FK6 [INFO] patches  -- 11/11 --  1048/4275 ( 24.5% ) -- 
2023-06-05 13:29:20 599-JY5FK6 [INFO] patches  -- 11/11 --  1088/4275 ( 25.5% ) -- 
2023-06-05 13:29:20 599-JY5FK6 [INFO] patches  -- 11/11 --  1128/4275 ( 26.4% ) -- 
2023-06-05 13:29:21 599-JY5FK6 [INFO] patches  -- 11/11 --  1168/4275 ( 27.3% ) -- 
2023-06-05 13:29:21 599-JY5FK6 [INFO] patches  -- 11/11 --  1208/4275 ( 28.3% ) -- 
2023-06-05 13:29:22 599-JY5FK6 [INFO] patches  -- 11/11 --  1248/4275 ( 29.2% ) -- 
2023-06-05 13:29:22 599-JY5FK6 [INFO] patches  -- 11/11 --  1288/4275 ( 30.1% ) -- 
2023-06-05 13:29:22 599-JY5FK6 [INFO] patches  -- 11/11 --  1328/4275 ( 31.1% ) -- 
2023-06-05 13:29:23 599-JY5FK6 [INFO] patches  -- 11/11 --  1368/4275 ( 32.0% ) -- 
2023-06-05 13:29:23 599-JY5FK6 [INFO] patches  -- 11/11 --  1408/4275 ( 32.9% ) -- 
2023-06-05 13:29:24 599-JY5FK6 [INFO] patches  -- 11/11 --  1448/4275 ( 33.9% ) -- 
2023-06-05 13:29:24 599-JY5FK6 [INFO] patches  -- 11/11 --  1488/4275 ( 34.8% ) -- 
2023-06-05 13:29:25 599-JY5FK6 [INFO] patches  -- 11/11 --  1528/4275 ( 35.7% ) -- 
2023-06-05 13:29:25 599-JY5FK6 [INFO] patches  -- 11/11 --  1568/4275 ( 36.7% ) -- 
2023-06-05 13:29:26 599-JY5FK6 [INFO] patches  -- 11/11 --  1608/4275 ( 37.6% ) -- 
2023-06-05 13:29:26 599-JY5FK6 [INFO] patches  -- 11/11 --  1648/4275 ( 38.5% ) -- 
2023-06-05 13:29:27 599-JY5FK6 [INFO] patches  -- 11/11 --  1688/4275 ( 39.5% ) -- 
2023-06-05 13:29:27 599-JY5FK6 [INFO] patches  -- 11/11 --  1728/4275 ( 40.4% ) -- 
2023-06-05 13:29:27 599-JY5FK6 [INFO] patches  -- 11/11 --  1768/4275 ( 41.4% ) -- 
2023-06-05 13:29:28 599-JY5FK6 [INFO] patches  -- 11/11 --  1808/4275 ( 42.3% ) -- 
2023-06-05 13:29:28 599-JY5FK6 [INFO] patches  -- 11/11 --  1848/4275 ( 43.2% ) -- 
2023-06-05 13:29:29 599-JY5FK6 [INFO] patches  -- 11/11 --  1888/4275 ( 44.2% ) -- 
2023-06-05 13:29:29 599-JY5FK6 [INFO] patches  -- 11/11 --  1928/4275 ( 45.1% ) -- 
2023-06-05 13:29:30 599-JY5FK6 [INFO] patches  -- 11/11 --  1968/4275 ( 46.0% ) -- 
2023-06-05 13:29:30 599-JY5FK6 [INFO] patches  -- 11/11 --  2008/4275 ( 47.0% ) -- 
2023-06-05 13:29:30 599-JY5FK6 [INFO] patches  -- 11/11 --  2048/4275 ( 47.9% ) -- 
2023-06-05 13:29:31 599-JY5FK6 [INFO] patches  -- 11/11 --  2088/4275 ( 48.8% ) -- 
2023-06-05 13:29:31 599-JY5FK6 [INFO] patches  -- 11/11 --  2128/4275 ( 49.8% ) -- 
2023-06-05 13:29:32 599-JY5FK6 [INFO] patches  -- 11/11 --  2168/4275 ( 50.7% ) -- 
2023-06-05 13:29:32 599-JY5FK6 [INFO] patches  -- 11/11 --  2208/4275 ( 51.6% ) -- 
2023-06-05 13:29:33 599-JY5FK6 [INFO] patches  -- 11/11 --  2248/4275 ( 52.6% ) -- 
2023-06-05 13:29:33 599-JY5FK6 [INFO] patches  -- 11/11 --  2288/4275 ( 53.5% ) -- 
2023-06-05 13:29:34 599-JY5FK6 [INFO] patches  -- 11/11 --  2328/4275 ( 54.5% ) -- 
2023-06-05 13:29:34 599-JY5FK6 [INFO] patches  -- 11/11 --  2368/4275 ( 55.4% ) -- 
2023-06-05 13:29:34 599-JY5FK6 [INFO] patches  -- 11/11 --  2408/4275 ( 56.3% ) -- 
2023-06-05 13:29:35 599-JY5FK6 [INFO] patches  -- 11/11 --  2448/4275 ( 57.3% ) -- 
2023-06-05 13:29:35 599-JY5FK6 [INFO] patches  -- 11/11 --  2488/4275 ( 58.2% ) -- 
2023-06-05 13:29:36 599-JY5FK6 [INFO] patches  -- 11/11 --  2528/4275 ( 59.1% ) -- 
2023-06-05 13:29:36 599-JY5FK6 [INFO] patches  -- 11/11 --  2568/4275 ( 60.1% ) -- 
2023-06-05 13:29:37 599-JY5FK6 [INFO] patches  -- 11/11 --  2608/4275 ( 61.0% ) -- 
2023-06-05 13:29:37 599-JY5FK6 [INFO] patches  -- 11/11 --  2648/4275 ( 61.9% ) -- 
2023-06-05 13:29:38 599-JY5FK6 [INFO] patches  -- 11/11 --  2688/4275 ( 62.9% ) -- 
2023-06-05 13:29:38 599-JY5FK6 [INFO] patches  -- 11/11 --  2728/4275 ( 63.8% ) -- 
2023-06-05 13:29:39 599-JY5FK6 [INFO] patches  -- 11/11 --  2768/4275 ( 64.7% ) -- 
2023-06-05 13:29:39 599-JY5FK6 [INFO] patches  -- 11/11 --  2808/4275 ( 65.7% ) -- 
2023-06-05 13:29:39 599-JY5FK6 [INFO] patches  -- 11/11 --  2848/4275 ( 66.6% ) -- 
2023-06-05 13:29:40 599-JY5FK6 [INFO] patches  -- 11/11 --  2888/4275 ( 67.6% ) -- 
2023-06-05 13:29:40 599-JY5FK6 [INFO] patches  -- 11/11 --  2928/4275 ( 68.5% ) -- 
2023-06-05 13:29:41 599-JY5FK6 [INFO] patches  -- 11/11 --  2968/4275 ( 69.4% ) -- 
2023-06-05 13:29:41 599-JY5FK6 [INFO] patches  -- 11/11 --  3008/4275 ( 70.4% ) -- 
2023-06-05 13:29:42 599-JY5FK6 [INFO] patches  -- 11/11 --  3048/4275 ( 71.3% ) -- 
2023-06-05 13:29:42 599-JY5FK6 [INFO] patches  -- 11/11 --  3088/4275 ( 72.2% ) -- 
2023-06-05 13:29:43 599-JY5FK6 [INFO] patches  -- 11/11 --  3128/4275 ( 73.2% ) -- 
2023-06-05 13:29:43 599-JY5FK6 [INFO] patches  -- 11/11 --  3168/4275 ( 74.1% ) -- 
2023-06-05 13:29:44 599-JY5FK6 [INFO] patches  -- 11/11 --  3208/4275 ( 75.0% ) -- 
2023-06-05 13:29:44 599-JY5FK6 [INFO] patches  -- 11/11 --  3248/4275 ( 76.0% ) -- 
2023-06-05 13:29:44 599-JY5FK6 [INFO] patches  -- 11/11 --  3288/4275 ( 76.9% ) -- 
2023-06-05 13:29:45 599-JY5FK6 [INFO] patches  -- 11/11 --  3328/4275 ( 77.8% ) -- 
2023-06-05 13:29:45 599-JY5FK6 [INFO] patches  -- 11/11 --  3368/4275 ( 78.8% ) -- 
2023-06-05 13:29:46 599-JY5FK6 [INFO] patches  -- 11/11 --  3408/4275 ( 79.7% ) -- 
2023-06-05 13:29:46 599-JY5FK6 [INFO] patches  -- 11/11 --  3448/4275 ( 80.7% ) -- 
2023-06-05 13:29:47 599-JY5FK6 [INFO] patches  -- 11/11 --  3488/4275 ( 81.6% ) -- 
2023-06-05 13:29:47 599-JY5FK6 [INFO] patches  -- 11/11 --  3528/4275 ( 82.5% ) -- 
2023-06-05 13:29:48 599-JY5FK6 [INFO] patches  -- 11/11 --  3568/4275 ( 83.5% ) -- 
2023-06-05 13:29:48 599-JY5FK6 [INFO] patches  -- 11/11 --  3608/4275 ( 84.4% ) -- 
2023-06-05 13:29:49 599-JY5FK6 [INFO] patches  -- 11/11 --  3648/4275 ( 85.3% ) -- 
2023-06-05 13:29:49 599-JY5FK6 [INFO] patches  -- 11/11 --  3688/4275 ( 86.3% ) -- 
2023-06-05 13:29:49 599-JY5FK6 [INFO] patches  -- 11/11 --  3728/4275 ( 87.2% ) -- 
2023-06-05 13:29:50 599-JY5FK6 [INFO] patches  -- 11/11 --  3768/4275 ( 88.1% ) -- 
2023-06-05 13:29:50 599-JY5FK6 [INFO] patches  -- 11/11 --  3808/4275 ( 89.1% ) -- 
2023-06-05 13:29:51 599-JY5FK6 [INFO] patches  -- 11/11 --  3848/4275 ( 90.0% ) -- 
2023-06-05 13:29:51 599-JY5FK6 [INFO] patches  -- 11/11 --  3888/4275 ( 90.9% ) -- 
2023-06-05 13:29:52 599-JY5FK6 [INFO] patches  -- 11/11 --  3928/4275 ( 91.9% ) -- 
2023-06-05 13:29:52 599-JY5FK6 [INFO] patches  -- 11/11 --  3968/4275 ( 92.8% ) -- 
2023-06-05 13:29:52 599-JY5FK6 [INFO] patches  -- 11/11 --  4008/4275 ( 93.8% ) -- 
2023-06-05 13:29:53 599-JY5FK6 [INFO] patches  -- 11/11 --  4048/4275 ( 94.7% ) -- 
2023-06-05 13:29:53 599-JY5FK6 [INFO] patches  -- 11/11 --  4088/4275 ( 95.6% ) -- 
2023-06-05 13:29:54 599-JY5FK6 [INFO] patches  -- 11/11 --  4128/4275 ( 96.6% ) -- 
2023-06-05 13:29:54 599-JY5FK6 [INFO] patches  -- 11/11 --  4168/4275 ( 97.5% ) -- 
2023-06-05 13:29:55 599-JY5FK6 [INFO] patches  -- 11/11 --  4208/4275 ( 98.4% ) -- 
2023-06-05 13:29:55 599-JY5FK6 [INFO] patches  -- 11/11 --  4248/4275 ( 99.4% ) -- 
[INFO] Total time: 0m 48s

[66]:

my_classifier.show_inference_sample_results(
    label="rail space", num_samples=9, set_name="patches", min_conf=80
)

../_images/Worked-examples_one_inch_pipeline_85_0.png

Add predictions to dataframe

[67]:

predictions_df = patch_dataset.patch_df

[68]:

import numpy as np
import pandas as pd

predictions_df["predicted_label"] = my_classifier.pred_label
predictions_df["pred"] = my_classifier.pred_label_indices
predictions_df["conf"] = np.array(my_classifier.pred_conf).max(axis=1)

predictions_df.head()

[68]:

	parent_id	image_path	pixel_bounds	shape	coordinates	crs	polygon	mean_pixel_R	mean_pixel_G	mean_pixel_B	mean_pixel_A	std_pixel_R	std_pixel_G	std_pixel_B	std_pixel_A	image_id	predicted_label	pred	conf
image_id
patch-4800-4700-4900-4800-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(4800, 4700, 4900, 4800)	(100, 100, 4)	(-4.436203529094827, 56.016994037098904, -4.42...	EPSG:4326	POLYGON ((-4.42791642813847 56.016994037098904...	0.708448	0.694029	0.628624	1.0	0.192450	0.194525	0.179188	0.0	patch-4800-4700-4900-4800-#map_74488689.png#.png	No	0	0.832045
patch-1700-200-1800-300-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(1700, 200, 1800, 300)	(100, 100, 4)	(-4.693103658741918, 55.81046025058123, -4.684...	EPSG:4326	POLYGON ((-4.6848165577855605 55.8104602505812...	0.902684	0.889986	0.817024	1.0	0.055550	0.056337	0.053893	0.0	patch-1700-200-1800-300-#map_74488689.png#.png	No	0	0.807598
patch-300-500-400-600-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(300, 500, 400, 600)	(100, 100, 4)	(-4.809123072130927, 55.82422916968241, -4.800...	EPSG:4326	POLYGON ((-4.800835971174569 55.82422916968241...	0.780055	0.768753	0.704215	1.0	0.118848	0.120624	0.114077	0.0	patch-300-500-400-600-#map_74488689.png#.png	No	0	0.953660
patch-2000-3500-2100-3600-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(2000, 3500, 2100, 3600)	(100, 100, 4)	(-4.668242355872845, 55.96191836069419, -4.659...	EPSG:4326	POLYGON ((-4.659955254916487 55.96191836069419...	0.835125	0.819675	0.744910	1.0	0.119352	0.119904	0.111668	0.0	patch-2000-3500-2100-3600-#map_74488689.png#.png	No	0	0.826236
patch-6200-4400-6300-4500-#map_74488689.png#.png	map_74488689.png	/Users/rwood/LwM/MapReader/worked_examples/geo...	(6200, 4400, 6300, 4500)	(100, 100, 4)	(-4.320184115705819, 56.00322511799772, -4.311...	EPSG:4326	POLYGON ((-4.311897014749461 56.00322511799772...	0.671154	0.655859	0.590913	1.0	0.213728	0.216932	0.199242	0.0	patch-6200-4400-6300-4500-#map_74488689.png#.png	rail space	1	0.995773

Add outputs to `MapImages` as metadata

[69]:

my_maps.add_metadata(patch_df, tree_level="patch")

[74]:

my_maps.add_shape()

[INFO] Add shape, tree level: parent

[75]:

parent_df, patch_df = my_maps.convert_images()
parent_df.head()

[75]:

	parent_id	image_path	patches	name	url	coordinates	crs	published_date	grid_bb	shape
image_id
map_74488689.png	None	/Users/rwood/LwM/MapReader/worked_examples/geo...	[patch-4800-4700-4900-4800-#map_74488689.png#....	map_74488689.png	https://maps.nls.uk/view/74488689	(-4.833984375, 55.80128097118045, -4.21875, 56...	EPSG:4326	1898	[(14, 7972, 5097)x(14, 8000, 5118)]	(5632, 7424, 4)

[76]:

parent_list = my_maps.list_parents()

my_maps.show_parent(
    parent_list[0],
    column_to_plot="pred",
    patch_border=True,
    plot_parent=True,
    figsize=(15, 15),
    alpha=0.5,
)

100%|█████████████████████████████████████████████████████████████████████████████| 1/1 [00:02<00:00,  2.48s/it]

[76]:

[<Figure size 1500x1500 with 2 Axes>]

../_images/Worked-examples_one_inch_pipeline_93_2.png

[77]:

parent_list = my_maps.list_parents()

my_maps.show_parent(
    parent_list[0],
    column_to_plot="conf",
    patch_border=True,
    plot_parent=True,
    figsize=(15, 15),
    alpha=0.5,
)

100%|█████████████████████████████████████████████████████████████████████████████| 1/1 [00:02<00:00,  2.51s/it]

[77]:

[<Figure size 1500x1500 with 2 Axes>]

../_images/Worked-examples_one_inch_pipeline_94_2.png

[ ]: