Add/Port granular video code processing

iscc_sdk/main.py

@@ -253,10 +253,27 @@ def code_video(fp, extract_meta=None, create_thumb=None):
             thumbnail_durl = idk.image_to_data_url(thumbnail_image)
             meta["thumbnail"] = thumbnail_durl
 
-    features = idk.video_features_extract(fp)
+    sig, scenes = None, []
+    if idk.sdk_opts.granular:
+        sig, scenes = idk.video_mp7sig_extract_scenes(fp)
+    else:
+        sig = idk.video_mp7sig_extract(fp)
+
+    if idk.sdk_opts.video_store_mp7sig:
+        outp = fp + ".iscc.mp7sig"
+        with open(outp, "wb") as outf:
+            outf.write(sig)
+
+    frames = idk.read_mp7_signature(sig)
+    features = [tuple(frame.vector.tolist()) for frame in frames]
+
     code_obj = ic.gen_video_code_v0(features, bits=idk.core_opts.video_bits)
     meta.update(code_obj)
 
+    if idk.sdk_opts.granular:
+        granular = idk.video_compute_granular(frames, scenes)
+        meta["features"] = granular
+
     return idk.IsccMeta.construct(**meta)
 
 

iscc_sdk/mp7.py

@@ -11,6 +11,7 @@
 
 __all__ = [
     "read_mp7_signature",
+    "Frame",
 ]
 
 

iscc_sdk/video.py

@@ -11,6 +11,7 @@
 from secrets import token_hex
 from PIL import Image, ImageEnhance
 import iscc_sdk as idk
+import iscc_core as ic
 
 
 __all__ = [
@@ -21,6 +22,7 @@
     "video_mp7sig_extract_scenes",
     "video_features_extract",
     "video_parse_scenes",
+    "video_compute_granular",
 ]
 
 VIDEO_META_MAP = {
@@ -325,3 +327,35 @@ def video_parse_scenes(scene_text, scene_limit=None):
         cutpoints.append(times[-1])
 
     return cutpoints[1:]
+
+
+def video_compute_granular(frames, scenes):
+    # type: (List[idk.Frame], List[float]) -> dict
+    """Compute video signatures for individual scenes in video.
+    Returns a dictionary conforming to `shema.Feature`- objects.
+    """
+    features, sizes, segment = [], [], []
+    start_frame = 0
+    for cidx, cutpoint in enumerate(scenes):
+        try:
+            frames = frames[start_frame:]
+        except IndexError:  # pragma: no cover
+            break
+        for fidx, frame in enumerate(frames):
+            frame_t = tuple(frame.vector.tolist())
+            segment.append(frame_t)
+            if frame.elapsed >= cutpoint:
+                features.append(ic.encode_base64(ic.soft_hash_video_v0(segment, 64)))
+                segment = []
+                prev_cutpoint = 0 if cidx == 0 else scenes[cidx - 1]
+                duration = round(cutpoint - prev_cutpoint, 3)
+                sizes.append(duration)
+                start_frame = fidx + 1
+                break
+    if not features:
+        log.info("No scenes detected. Use all frames")
+        segment = [tuple(frame.vector.tolist()) for frame in frames]
+        features = [ic.encode_base64(ic.soft_hash_video_v0(segment, 64))]
+        sizes = [round(float(frames[-1].elapsed), 3)]
+
+    return dict(kind="video", version=0, features=features, sizes=sizes)

tests/test_video.py

@@ -114,3 +114,89 @@ def test_code_video_nometa_nothumb(mp4_file):
     assert meta.dict() == {"iscc": "ISCC:EMAV4DUD6QORW4X4"}
     idk.sdk_opts.extract_metadata = True
     idk.sdk_opts.create_thumbnail = True
+
+
+def test_code_video_granular_scenes(mp4_file):
+    idk.sdk_opts.granular = True
+    idk.sdk_opts.video_scene_limit = 0.2
+    idk.sdk_opts.create_thumbnail = False
+    idk.sdk_opts.extract_metadata = True
+    result = idk.code_video(mp4_file).dict()
+    assert result == {
+        "iscc": "ISCC:EMAV4DUD6QORW4X4",
+        "name": "Kali by Anokato - Spiral Sessions 2019",
+        "features": {
+            "kind": "video",
+            "version": 0,
+            "sizes": [7.625, 2.5, 5.083, 20.792, 2.458, 1.5, 6.667],
+            "features": [
+                "XxqT9x1acvw",
+                "HEqSawAW8oQ",
+                "VA7Q9A0esuw",
+                "Xg4H9H1S8vU",
+                "l06Qp9wbcow",
+                "UWAQkpxZMqg",
+                "HloAnYYVUqU",
+            ],
+        },
+    }
+
+
+def test_code_iscc_video_granular(mp4_file):
+    idk.sdk_opts.granular = True
+    idk.sdk_opts.video_scene_limit = 0.2
+    idk.sdk_opts.create_thumbnail = False
+    idk.sdk_opts.extract_metadata = True
+    result = idk.code_iscc(mp4_file).dict()
+    assert result == {
+        "@type": "VideoObject",
+        "iscc": "ISCC:KMCV6UK6BSXJ3I4GLYHIH5A5DNZPYBWQO33FNHPQFOOUCLLW3HKRNUA",
+        "name": "Kali by Anokato - Spiral Sessions 2019",
+        "mode": "video",
+        "mediatype": "video/mp4",
+        "filename": "video.mp4",
+        "filesize": 2161914,
+        "metahash": "1e2096c0a53475a186ce37622aba7ba70651fc62cc8150f59eee6d17dc16d9bfbf25",
+        "datahash": "1e209d412d76d9d516d07bb60f1ab3c1a5c1b176ed4f1cec94c96222a5d013ec3e38",
+        "features": {
+            "kind": "video",
+            "version": 0,
+            "features": [
+                "XxqT9x1acvw",
+                "HEqSawAW8oQ",
+                "VA7Q9A0esuw",
+                "Xg4H9H1S8vU",
+                "l06Qp9wbcow",
+                "UWAQkpxZMqg",
+                "HloAnYYVUqU",
+            ],
+            "sizes": [7.625, 2.5, 5.083, 20.792, 2.458, 1.5, 6.667],
+        },
+    }
+
+
+def test_code_iscc_video_granular_no_scenes(mp4_file):
+    idk.sdk_opts.granular = True
+    idk.sdk_opts.video_scene_limit = 0.8
+    idk.sdk_opts.create_thumbnail = False
+    idk.sdk_opts.extract_metadata = True
+    result = idk.code_iscc(mp4_file).dict()
+    assert result == {
+        "@type": "VideoObject",
+        "iscc": "ISCC:KMCV6UK6BSXJ3I4GLYHIH5A5DNZPYBWQO33FNHPQFOOUCLLW3HKRNUA",
+        "name": "Kali by Anokato - Spiral Sessions 2019",
+        "mode": "video",
+        "mediatype": "video/mp4",
+        "filename": "video.mp4",
+        "filesize": 2161914,
+        "metahash": "1e2096c0a53475a186ce37622aba7ba70651fc62cc8150f59eee6d17dc16d9bfbf25",
+        "datahash": "1e209d412d76d9d516d07bb60f1ab3c1a5c1b176ed4f1cec94c96222a5d013ec3e38",
+        "features": {
+            "kind": "video",
+            "version": 0,
+            "features": [
+                "Xg6D9B0bcvw",
+            ],
+            "sizes": [59.8],
+        },
+    }