Slicing with SAHI¶

0. Preparation¶

• Be sure current working directory is 'demo/' folder:

In [ ]:

import os
os.getcwd()

import os os.getcwd()

• Import required functions:

In [1]:

from sahi.slicing import slice_coco
from sahi.utils.file import load_json

from PIL import Image, ImageDraw
import matplotlib.pyplot as plt

from sahi.slicing import slice_coco from sahi.utils.file import load_json from PIL import Image, ImageDraw import matplotlib.pyplot as plt

1. Slicing COCO Dataset into Grids¶

• Visualize original image and annotations:

In [2]:

coco_dict = load_json("demo_data/terrain2_coco.json")

f, axarr = plt.subplots(1, 1, figsize=(12, 12))
# read image
img_ind = 0
img = Image.open("demo_data/" + coco_dict["images"][img_ind]["file_name"]).convert('RGBA')
# iterate over all annotations
for ann_ind in range(len(coco_dict["annotations"])):
    # convert coco bbox to pil bbox
    xywh = coco_dict["annotations"][ann_ind]["bbox"]
    xyxy = [xywh[0], xywh[1], xywh[0]+xywh[2], xywh[1]+xywh[3]]
    # visualize bbox over image
    ImageDraw.Draw(img, 'RGBA').rectangle(xyxy, width=5)
axarr.imshow(img)

coco_dict = load_json("demo_data/terrain2_coco.json") f, axarr = plt.subplots(1, 1, figsize=(12, 12)) # read image img_ind = 0 img = Image.open("demo_data/" + coco_dict["images"][img_ind]["file_name"]).convert('RGBA') # iterate over all annotations for ann_ind in range(len(coco_dict["annotations"])): # convert coco bbox to pil bbox xywh = coco_dict["annotations"][ann_ind]["bbox"] xyxy = [xywh[0], xywh[1], xywh[0]+xywh[2], xywh[1]+xywh[3]] # visualize bbox over image ImageDraw.Draw(img, 'RGBA').rectangle(xyxy, width=5) axarr.imshow(img)

Out[2]:

<matplotlib.image.AxesImage at 0x7fdb3fe44310>

• To slice a COCO dataset annoations an images, we need to specify slice parameters. In this example we will slice images into 256x256 grids overlap ratio of 0.2:

In [3]:

coco_dict, coco_path = slice_coco(
    coco_annotation_file_path="demo_data/terrain2_coco.json",
    image_dir="demo_data/",
    output_coco_annotation_file_name="sliced_coco.json",
    ignore_negative_samples=False,
    output_dir="demo_data/sliced/",
    slice_height=256,
    slice_width=256,
    overlap_height_ratio=0.2,
    overlap_width_ratio=0.2,
    min_area_ratio=0.1,
    verbose=True
)

coco_dict, coco_path = slice_coco( coco_annotation_file_path="demo_data/terrain2_coco.json", image_dir="demo_data/", output_coco_annotation_file_name="sliced_coco.json", ignore_negative_samples=False, output_dir="demo_data/sliced/", slice_height=256, slice_width=256, overlap_height_ratio=0.2, overlap_width_ratio=0.2, min_area_ratio=0.1, verbose=True )

100%|██████████| 5/5 [00:00<00:00, 121222.66it/s]
100%|██████████| 1/1 [00:00<00:00, 10.75it/s]

indexing coco dataset annotations...
image.shape: (1024, 682)
sliced image path: demo_data/sliced/terrain2_0_0_256_256.png
sliced image path: demo_data/sliced/terrain2_205_0_461_256.png
sliced image path: demo_data/sliced/terrain2_410_0_666_256.png
sliced image path: demo_data/sliced/terrain2_615_0_871_256.png
sliced image path: demo_data/sliced/terrain2_768_0_1024_256.png
sliced image path: demo_data/sliced/terrain2_0_205_256_461.png
sliced image path: demo_data/sliced/terrain2_205_205_461_461.png
sliced image path: demo_data/sliced/terrain2_410_205_666_461.png
sliced image path: demo_data/sliced/terrain2_615_205_871_461.png
sliced image path: demo_data/sliced/terrain2_768_205_1024_461.png
sliced image path: demo_data/sliced/terrain2_0_410_256_666.png
sliced image path: demo_data/sliced/terrain2_205_410_461_666.png
sliced image path: demo_data/sliced/terrain2_410_410_666_666.png
sliced image path: demo_data/sliced/terrain2_615_410_871_666.png
sliced image path: demo_data/sliced/terrain2_768_410_1024_666.png
sliced image path: demo_data/sliced/terrain2_0_426_256_682.png
sliced image path: demo_data/sliced/terrain2_205_426_461_682.png
sliced image path: demo_data/sliced/terrain2_410_426_666_682.png
sliced image path: demo_data/sliced/terrain2_615_426_871_682.png
sliced image path: demo_data/sliced/terrain2_768_426_1024_682.png
Num slices: 20 slice_height 256 slice_width 256
Time to slice demo_data/terrain2.png 0.09270477294921875 seconds

• Visualize sliced images:

In [4]:

f, axarr = plt.subplots(4, 5, figsize=(13,13))
img_ind = 0
for ind1 in range(4):
    for ind2 in range(5):
        img = Image.open("demo_data/sliced/" + coco_dict["images"][img_ind]["file_name"])
        axarr[ind1, ind2].imshow(img)
        img_ind += 1

f, axarr = plt.subplots(4, 5, figsize=(13,13)) img_ind = 0 for ind1 in range(4): for ind2 in range(5): img = Image.open("demo_data/sliced/" + coco_dict["images"][img_ind]["file_name"]) axarr[ind1, ind2].imshow(img) img_ind += 1

• Visualize sliced annotations on sliced images:

In [5]:

f, axarr = plt.subplots(4, 5, figsize=(13,13))
img_ind = 0
for row_ind in range(4):
    for column_ind in range(5):
        # read image
        img = Image.open("demo_data/sliced/" + coco_dict["images"][img_ind]["file_name"]).convert('RGBA')
        # iterate over all annotations
        for ann_ind in range(len(coco_dict["annotations"])):
            # find annotations that belong the selected image
            if coco_dict["annotations"][ann_ind]["image_id"] == coco_dict["images"][img_ind]["id"]:
                # convert coco bbox to pil bbox
                xywh = coco_dict["annotations"][ann_ind]["bbox"]
                xyxy = [xywh[0], xywh[1], xywh[0]+xywh[2], xywh[1]+xywh[3]]
                # visualize bbox over image
                ImageDraw.Draw(img, 'RGBA').rectangle(xyxy, width=5)
            axarr[row_ind, column_ind].imshow(img)
        img_ind += 1

f, axarr = plt.subplots(4, 5, figsize=(13,13)) img_ind = 0 for row_ind in range(4): for column_ind in range(5): # read image img = Image.open("demo_data/sliced/" + coco_dict["images"][img_ind]["file_name"]).convert('RGBA') # iterate over all annotations for ann_ind in range(len(coco_dict["annotations"])): # find annotations that belong the selected image if coco_dict["annotations"][ann_ind]["image_id"] == coco_dict["images"][img_ind]["id"]: # convert coco bbox to pil bbox xywh = coco_dict["annotations"][ann_ind]["bbox"] xyxy = [xywh[0], xywh[1], xywh[0]+xywh[2], xywh[1]+xywh[3]] # visualize bbox over image ImageDraw.Draw(img, 'RGBA').rectangle(xyxy, width=5) axarr[row_ind, column_ind].imshow(img) img_ind += 1