T-SQL for create and populate an image table; stored procedure to load pre-trained Caffe2 model to detect places.

samples/CaffeModelwithSQLPython/CreateTable.sql

@@ -0,0 +1,17 @@
+DROP TABLE [dbo].[PlaceImages]
+
+SET ANSI_NULLS ON
+GO
+
+SET QUOTED_IDENTIFIER ON
+GO
+
+CREATE TABLE [dbo].[PlaceImages](
+	[ImageID] [int] IDENTITY(1,1) NOT NULL,
+	[ImagePath] [nvarchar](250) NOT NULL,
+	[ImageRaw] [varbinary](max) NULL,
+	[PlaceDetected] [varchar](250) NULL,
+	[Probability] [float] NULL
+)
+
+GO

samples/CaffeModelwithSQLPython/sp_place_detection.sql

@@ -0,0 +1,94 @@
+CREATE PROCEDURE [dbo].[usp_ImagesDB_detect_places]
+AS
+BEGIN
+DECLARE @MyTempTable TABLE
+    (
+        ImageID   int NOT NULL,
+        Place   varchar(50),
+		Prob    float NULL
+    );
+
+INSERT INTO @MyTempTable
+EXECUTE sp_execute_external_script
+@language=N'Python',
+@script=N'
+import pandas as pd
+# load up the caffe2 workspace
+from caffe2.python import workspace
+from caffe2.python.tutorials import helpers
+
+def map_results(results):
+    import numpy as np
+    import csv
+    results = np.asarray(results)
+    results = np.delete(results, 1)
+    index = 0
+    highest = 0
+    arr = np.empty((0,2), dtype=object)
+    arr[:,0] = int(10)
+    arr[:,1:] = float(10)
+    for i, r in enumerate(results):
+        # imagenet index begins with 1!
+        i=i+1
+        arr = np.append(arr, np.array([[i,r]]), axis=0)
+        if (r > highest):
+            highest = r
+            index = i
+
+    # top 3 results
+    #print("Raw top 3 results:", sorted(arr, key=lambda x: x[1], reverse=True)[:3])
+
+    CATEGORY_MAPPING = "C:\demo\models\places-cnn\categories_places365.txt"
+    with open(CATEGORY_MAPPING,''r'') as f:
+        for line in f:
+            categoryname, label = line.partition(" ")[::2]
+            if (label.strip() == str(index)):
+                category_result = categoryname.strip().split("/")[2]
+                prob = highest
+    return category_result, prob
+
+# load the pre-trained model
+# the Caffe2 models here was converted from Places365 published by http://places.csail.mit.edu/
+CAFFE_MODELS = "C:\demo\models"
+MODEL = ''places-cnn'', ''init_net.pb'', ''predict_net.pb''
+INIT_NET = os.path.join(CAFFE_MODELS, MODEL[0], MODEL[1])
+PREDICT_NET = os.path.join(CAFFE_MODELS, MODEL[0], MODEL[2])
+with open(INIT_NET, ''rb'') as f:
+    init_net = f.read()
+with open(PREDICT_NET, ''rb'') as f:
+    predict_net = f.read()
+
+p = workspace.Predictor(init_net, predict_net)
+
+predictions = [''''] * len(InputDataSet)
+probs = [0] * len(InputDataSet)
+for i in range(0, len(InputDataSet)):
+    img = InputDataSet.iloc[i][''ImagePath'']
+
+	# dimension of the images that the model was trained with
+    input_dim = 227
+    mean = 128
+
+    # use the image helper to load the image and perform image pre-processing
+    img = helpers.loadToNCHW(img, mean, input_dim)
+
+    # submit the image to net and get a tensor of results
+    results = p.run([img]) 
+
+    # lookup the result from the places category list
+    predictions[i], probs[i] = map_results(results)
+
+InputDataSet.drop([''ImagePath''], axis=1, inplace=True)
+InputDataSet[''Place''] = predictions
+InputDataSet[''Prob''] = probs
+OutputDataSet = InputDataSet
+',      
+@input_data_1 = N'SELECT ImageID, ImagePath FROM dbo.PlaceImages WHERE PlaceDetected is null'    
+;   
+	  
+UPDATE dbo.PlaceImages 
+SET PlaceDetected = tb.Place, Probability = tb.Prob
+FROM @MyTempTable AS tb
+WHERE tb.ImageID = dbo.PlaceImages.ImageID
+
+END