diff --git a/keyframe_extraction.py b/keyframe_extraction.py
deleted file mode 100644
index 45b0dc7..0000000
--- a/keyframe_extraction.py
+++ /dev/null
@@ -1,90 +0,0 @@
-import cv2  # type: ignore
-import os
-import logging
-from time import time
-from typing import List
-
-from dane.provenance import Provenance
-
-
-logger = logging.getLogger(__name__)
-
-
-# Step 2: run keyframe extraction based on list of keyframe indices
-def run(
-    input_file_path: str,
-    keyframe_indices: List[int],
-    keyframe_timestamps: List[int],
-    output_dir: str,
-) -> Provenance:
-    start_time = time()
-    logger.info("Extracting keyframe images now.")
-
-    keyframe_files = extract_keyframes(
-        media_file=input_file_path,
-        keyframe_indices=keyframe_indices,
-        keyframe_timestamps=keyframe_timestamps,
-        out_dir=output_dir,
-    )
-    logger.info(f"Extracted {len(keyframe_indices)} keyframes.")
-    return Provenance(
-        activity_name="Keyframe extraction",
-        activity_description="Extract keyframes (images) for listed frame indices",
-        start_time_unix=start_time,
-        processing_time_ms=time() - start_time,
-        input_data={
-            "input_file_path": input_file_path,
-            "keyframe_indices": str(keyframe_indices),
-        },
-        output_data={"Keyframe files": str(keyframe_files)},
-    )
-
-
-def extract_keyframes(
-    media_file: str,
-    keyframe_indices: list[int],
-    keyframe_timestamps: list[int],
-    out_dir: str,
-) -> list[str]:
-    if not os.path.exists(media_file):
-        raise IOError("Input video not found")
-
-    vcap = cv2.VideoCapture(media_file)
-    if not vcap.isOpened():
-        raise IOError("Unable to open video file " + media_file)
-
-    next_i = int(vcap.get(cv2.CAP_PROP_POS_FRAMES))
-    max_i = max(keyframe_indices)
-    fns = []
-    while next_i <= max_i:
-        ret = vcap.grab()
-        # FIXME e.g. index 13048 --> 521920, round makes this 521919
-        if next_i in keyframe_indices:
-            # timestamp = round(vcap.get(cv2.CAP_PROP_POS_MSEC))  # msec position
-            timestamp = keyframe_timestamps[keyframe_indices.index(int(next_i))]
-            logger.info(f"index {int(next_i)} = {timestamp}")
-            if ret:
-                _, frame = vcap.retrieve()
-                fn = os.path.join(out_dir, f"{timestamp}.jpg")
-                cv2.imwrite(fn, frame)
-                fns.append(fn)
-            else:
-                raise IOError(f"Unable to read keyframe {next_i}")
-        next_i = int(vcap.get(cv2.CAP_PROP_POS_FRAMES))
-    vcap.release()
-    return fns
-
-
-def get_fps(media_file: str):
-    logger.info(f"Getting FPS of: {media_file}")
-    return cv2.VideoCapture(media_file).get(cv2.CAP_PROP_FPS)
-
-
-def get_framecount(media_file: str):
-    logger.info(f"Getting frame count of: {media_file}")
-    return cv2.VideoCapture(media_file).get(cv2.CAP_PROP_FRAME_COUNT)
-
-
-def frame_index_to_timecode(frame_index: int, fps: float, out_format="ms"):
-    if out_format == "ms":
-        return round(frame_index / fps * 1000)