model과 백엔드 연결

@@ -11,12 +11,15 @@ from django.http import JsonResponse
 from django.shortcuts import get_object_or_404, render
 from api.models import Video, VideoFile
 from api.serializers import VideoSerializer, VideoFileSerializer
-from api import  file_upload_path
+from api import file_upload_path 
 
 import subprocess
 import shlex
 import json
 # Create your views here.
+import sys
+sys.path.insert(0, "/home/jun/documents/univ/PKH_Project1/web/backend/yt8m/esot3ria")
+import inference_pb
 
 def with_ffprobe(filename):
 
@@ -75,11 +78,14 @@ class VideoFileUploadView(APIView):
             # 동영상 길이 출력  
             runTime = with_ffprobe('/'+file_serializer.data['file_save_name'])
             print(runTime)
-
+            print(threshold)
             process = subprocess.Popen(['./runMediaPipe.sh %s %s' %(file_serializer.data['file_save_name'],runTime,)], shell = True)
             process.wait()
 
-            return Response(True, status=status.HTTP_201_CREATED)
+
+            result =  inference_pb.inference_pb('/tmp/mediapipe/features.pb', threshold)
+
+            return Response(result, status=status.HTTP_201_CREATED)
         else:
             return Response(file_serializer.errors, status=status.HTTP_400_BAD_REQUEST)
 

@@ -3,16 +3,17 @@ import tensorflow as tf
 from tensorflow import logging
 from tensorflow import gfile
 import operator
-import esot3ria.pb_util as pbutil
-import esot3ria.video_recommender as recommender
-import esot3ria.video_util as videoutil
+import pb_util as pbutil
+import video_recommender as recommender
+import video_util as videoutil
 
 # Define file paths.
-MODEL_PATH = "/Users/esot3ria/PycharmProjects/yt8m/models/frame/refined_model/inference_model/segment_inference_model"
-VOCAB_PATH = "../vocabulary.csv"
-VIDEO_TAGS_PATH = "./kaggle_solution_40k.csv"
-TAG_VECTOR_MODEL_PATH = "./tag_vectors.model"
-VIDEO_VECTOR_MODEL_PATH = "./video_vectors.model"
+MODEL_PATH = "/home/jun/documents/univ/PKH_Project1/web/backend/yt8m/esot3ria/model/inference_model/segment_inference_model"
+VOCAB_PATH = "/home/jun/documents/univ/PKH_Project1/web/backend/yt8m/vocabulary.csv"
+VIDEO_TAGS_PATH = "/home/jun/documents/univ/PKH_Project1/web/backend/yt8m/esot3ria/kaggle_solution_40k.csv"
+TAG_VECTOR_MODEL_PATH = "/home/jun/documents/univ/PKH_Project1/web/backend/yt8m/esot3ria/tag_vectors.model"
+VIDEO_VECTOR_MODEL_PATH = "/home/jun/documents/univ/PKH_Project1/web/backend/yt8m/esot3ria/video_vectors.model"
+SEGMENT_LABEL_PATH = "/home/jun/documents/univ/PKH_Project1/web/backend/yt8m/segment_label_ids.csv"
 
 # Define parameters.
 TAG_TOP_K = 5
@@ -83,7 +84,8 @@ def normalize_tag(tag):
         return tag
 
 
-def inference_pb(file_path):
+def inference_pb(file_path, threshold):
+    VIDEO_TOP_K = int(threshold)
     inference_result = {}
     with tf.Session(config=tf.ConfigProto(allow_soft_placement=True)) as sess:
 
@@ -135,8 +137,7 @@ def inference_pb(file_path):
 
         whitelisted_cls_mask = np.zeros((predictions_tensor.get_shape()[-1],),
                                         dtype=np.float32)
-        segment_label_ids_file = '../segment_label_ids.csv'
-        with tf.io.gfile.GFile(segment_label_ids_file) as fobj:
+        with tf.io.gfile.GFile(SEGMENT_LABEL_PATH) as fobj:
             for line in fobj:
                 try:
                     cls_id = int(line)
@@ -217,6 +218,6 @@ def inference_pb(file_path):
 
 
 if __name__ == '__main__':
-    filepath = "../featuremaps/features(yorusika).pb"
+    filepath = "/tmp/mediapipe/features.pb"
     result = inference_pb(filepath)
     print(result)

+model_checkpoint_path: "/root/volume/youtube-8m/saved_model/inference_model/segment_inference_model"
+all_model_checkpoint_paths: "/root/volume/youtube-8m/saved_model/inference_model/segment_inference_model"

+{"model": "FrameLevelLogisticModel", "feature_sizes": "1024,128", "feature_names": "rgb,audio", "frame_features": true, "label_loss": "CrossEntropyLoss"}
\ No newline at end of file