Update main_process/main_router.py

main_process/main_router.py

@@ -36,59 +36,6 @@ def get_casting(video_sha1: str):
 
     return faces_json, voices_json
 
-def cluster_secuencial_ocr(ocr_list, threshold=0.3):
-    if not ocr_list:
-        return []
-
-    ocr_text = [item["ocr"] for item in ocr_list if item and isinstance(item["ocr"], str)]
-    if not ocr_text:
-        return []
-
-    print(ocr_text)
-
-    model = SentenceTransformer("all-MiniLM-L6-v2")
-    embeddings = model.encode(ocr_text, normalize_embeddings=True)
-
-    clusters_representantes = []
-    if not embeddings.any():
-        return []
-
-    prev_embedding = embeddings[0]
-    start_time = ocr_list[0]["start"]
-    ocr_prev=ocr_text[0]
-
-    for i, emb in enumerate(embeddings[1:], 1):
-        ocr_actual=ocr_text[i]
-        sim = cosine_similarity([prev_embedding], [emb])[0][0]
-
-        print(ocr_prev, " - ", ocr_actual)
-        print(f"Similitud entre: {sim}")
-
-        if sim < threshold:
-            clusters_representantes.append({'index': i-1, 'start_time': start_time})
-            prev_embedding = emb
-            ocr_prev=ocr_actual
-            start_time = ocr_list[i]["start"]
-
-    clusters_representantes.append({'index': len(embeddings)-1, 'start_time': start_time})
-    print(clusters_representantes)
-
-    ocr_final = []
-    for cluster_info in clusters_representantes:
-        idx = cluster_info['index']
-
-        if idx < len(ocr_list) and ocr_list[idx]["ocr"]:
-            ocr_item = {
-                "ocr": ocr_list[idx]["ocr"],
-                "image_path": ocr_list[idx]["image_path"],
-                "start": cluster_info['start_time'],
-                "end": ocr_list[idx]["end"],
-                "faces": ocr_list[idx]["faces"]
-            }
-            ocr_final.append(ocr_item)
-
-    return ocr_final
-
 def map_identities_per_second(frames_per_second, intervals):
     for seg in intervals:
         seg_start = seg["start"]