Skip to content

Commit 03d5ec5

Browse files
zhong-alegrace479EmersonFras
authored
Update dataframe creation (#96)
* Rewrite dataframe creation * Fix merge * Remove unused DataFrame initialization --------- Co-authored-by: Elizabeth Campolongo <38985481+egrace479@users.noreply.github.com> Co-authored-by: Emersonfras <emersonfrasure@gmail.com>
1 parent 67493f0 commit 03d5ec5

File tree

1 file changed

+11
-11
lines changed

1 file changed

+11
-11
lines changed

src/kabr_tools/cvat2slowfast.py

Lines changed: 11 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -34,9 +34,9 @@ def cvat2slowfast(path_to_mini_scenes: str, path_to_new_dataset: str,
3434
with open(f"{path_to_new_dataset}/annotation/classes.json", "w") as file:
3535
json.dump(label2number, file)
3636

37-
headers = {"original_vido_id": [], "video_id": pd.Series(dtype="int"), "frame_id": pd.Series(dtype="int"),
38-
"path": [], "labels": []}
39-
charades_df = pd.DataFrame(data=headers)
37+
headers = ["original_vido_id", "video_id", "frame_id", "path", "labels"]
38+
charades_data = []
39+
4040
video_id = 1
4141
folder_name = 1
4242

@@ -127,13 +127,11 @@ def cvat2slowfast(path_to_mini_scenes: str, path_to_new_dataset: str,
127127
if not no_images:
128128
cv2.imwrite(f"{output_folder}/{adjusted_index}.jpg", frame)
129129

130-
# TODO: Major slow down here. Add to a list rather than dataframe,
131-
# and create dataframe at the end.
132-
charades_df.loc[len(charades_df.index)] = [f"{folder_code}",
133-
video_id,
134-
adjusted_index,
135-
f"{folder_code}/{adjusted_index}.jpg",
136-
str(label2number[behavior])]
130+
charades_data.append([f"{folder_code}",
131+
video_id,
132+
adjusted_index,
133+
f"{folder_code}/{adjusted_index}.jpg",
134+
str(label2number[behavior])])
137135

138136
adjusted_index += 1
139137

@@ -145,9 +143,11 @@ def cvat2slowfast(path_to_mini_scenes: str, path_to_new_dataset: str,
145143
video_id += 1
146144

147145
if video_id % 10 == 0:
146+
charades_df = pd.DataFrame(charades_data, columns=headers)
148147
charades_df.to_csv(
149148
f"{path_to_new_dataset}/annotation/data.csv", sep=" ", index=False)
150149

150+
charades_df = pd.DataFrame(charades_data, columns=headers)
151151
charades_df.to_csv(
152152
f"{path_to_new_dataset}/annotation/data.csv", sep=" ", index=False)
153153

@@ -203,4 +203,4 @@ def main() -> None:
203203

204204

205205
if __name__ == "__main__":
206-
main()
206+
main()

0 commit comments

Comments
 (0)