metadata
dataset_info:
features:
- name: image
dtype: image
- name: outlines
sequence:
sequence: uint8
- name: segments
sequence:
sequence:
sequence: bool
- name: id
dtype: int64
- name: is_bad
dtype: bool
splits:
- name: train
num_bytes: 18811729165.83
num_examples: 16455
- name: val
num_bytes: 3671501400.7200003
num_examples: 3209
- name: test
num_bytes: 3602703436.986
num_examples: 3138
download_size: 4189102455
dataset_size: 26085934003.536003
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: val
path: data/val-*
- split: test
path: data/test-*
Processed data from the Soccernet 2023 dataset. Processing notebook is included in this repo.
To see an example:
def show_item(item):
fig, axs = plt.subplots(nrows = 1, ncols = 4, figsize = (20, 4))
axs[0].imshow(item['image'])
axs[0].set_title("Image")
axs[0].axis('off')
axs[1].imshow(overlay_mask(item['image'], item['outlines']))
axs[1].set_title("Outlines")
axs[1].axis('off')
axs[2].imshow(show_segments(item['segments']))
axs[2].set_title("Segments")
axs[2].axis('off')
# PART 3: GET MASK OUTLINES
kernel = np.array([[0, 1, 0],
[1, -4, 1],
[0, 1, 0]])
segments = np.array(item['segments']).astype(np.uint8)
class_edges = np.zeros(segments.shape[1:], dtype=int)
for i in range(segments.shape[0]):
edge = convolve(segments[i], kernel, mode='constant', cval=0)
edge_detected = edge != 0
class_edges[edge_detected] = i
axs[3].imshow(overlay_mask(item['image'], class_edges))
axs[3].set_title("Segments Outlines")
axs[3].axis('off')
if item['is_bad']:
s = f"Bad ID: {item['id']}"
else:
s = f"ID: {item['id']}"
fig.suptitle(s, fontsize = 8)
plt.subplots_adjust(hspace = -0.2, wspace = -0.05)
plt.show()
show_item(dataset['train'][99])