Merge pull request robmarkcole#31 from robmarkcole/small-tidy

Small tidy

Author: robmarkcole

development/development-mobilenet_v2_coco.ipynb

@@ -1,9 +1,28 @@
-import requests
+"""Perform processing of test images."""
+import pprint
 
-API_URL = "http://localhost:5000/v1/vision/detection"
+import requests
 
-image_data = open("people_car.jpg", "rb").read()
+print("-----------------------------")
+OBJ_DETECTION_URL = "http://localhost:5000/v1/vision/detection"
+OBJ_DETECTION_TEST_IMAGE = "people_car.jpg"
+image_data = open(OBJ_DETECTION_TEST_IMAGE, "rb").read()
+response = requests.post(OBJ_DETECTION_URL, files={"image": image_data}).json()
+print(OBJ_DETECTION_TEST_IMAGE)
+pprint.pprint(response)
 
-response = requests.post(API_URL, files={"image": image_data}).json()
+print("-----------------------------")
+FACE_DETECTION_URL = "http://localhost:5000/v1/vision/face"
+FACE_DETECTION_TEST_IMAGE = "faces.jpg"
+image_data = open(FACE_DETECTION_TEST_IMAGE, "rb").read()
+response = requests.post(FACE_DETECTION_URL, files={"image": image_data}).json()
+print(FACE_DETECTION_TEST_IMAGE)
+pprint.pprint(response)
 
-print(response)
+print("-----------------------------")
+SCENE_URL = "http://localhost:5000/v1/vision/scene"
+SCENE_TEST_IMAGE = "dog.jpg"
+image_data = open(SCENE_TEST_IMAGE, "rb").read()
+response = requests.post(SCENE_URL, files={"image": image_data}).json()
+print(SCENE_TEST_IMAGE)
+pprint.pprint(response)

development/development-mobilenet_v2_face.ipynb

@@ -13,16 +13,16 @@
 
 app = FastAPI()
 
+# Settings
 MIN_CONFIDENCE = 0.1  # The absolute lowest confidence for a detection.
-
+# URL
 FACE_DETECTION_URL = "/v1/vision/face"
-FACE_MODEL = "models/face_detection/mobilenet_ssd_v2_face/mobilenet_ssd_v2_face_quant_postprocess.tflite"
-
 OBJ_DETECTION_URL = "/v1/vision/detection"
+SCENE_URL = "/v1/vision/scene"
+# Models and labels
+FACE_MODEL = "models/face_detection/mobilenet_ssd_v2_face/mobilenet_ssd_v2_face_quant_postprocess.tflite"
 OBJ_MODEL = "models/object_detection/mobilenet_ssd_v2_coco/mobilenet_ssd_v2_coco_quant_postprocess.tflite"
 OBJ_LABELS = "models/object_detection/mobilenet_ssd_v2_coco/coco_labels.txt"
-
-SCENE_URL = "/v1/vision/scene"
 SCENE_MODEL = "models/classification/dogs-vs-cats/model.tflite"
 SCENE_LABELS = "models/classification/dogs-vs-cats/labels.txt"
 
@@ -40,8 +40,8 @@
 face_interpreter.allocate_tensors()
 face_input_details = face_interpreter.get_input_details()
 face_output_details = face_interpreter.get_output_details()
-face_input_height = face_input_details[0]["shape"][1]  # 320
-face_input_width = face_input_details[0]["shape"][2]  # 320
+face_input_height = face_input_details[0]["shape"][1]
+face_input_width = face_input_details[0]["shape"][2]
 
 # Setup face detection
 scene_interpreter = tflite.Interpreter(model_path=SCENE_MODEL)
@@ -55,24 +55,19 @@
 
 @app.get("/")
 async def info():
-    return f"""
-        Object detection model: {OBJ_MODEL.split("/")[-2]}
-        Face detection model: {FACE_MODEL.split("/")[-2]}
-        Scene model: {SCENE_MODEL.split("/")[-2]}
-        """
+    return """tflite-server docs at ip:port/docs"""
 
 
 @app.post(FACE_DETECTION_URL)
 async def predict_face(image: UploadFile = File(...)):
-    data = {"success": False}
     try:
         contents = await image.read()
-        image = Image.open(io.BytesIO(contents))  # A PIL image
+        image = Image.open(io.BytesIO(contents))
         image_width = image.size[0]
         image_height = image.size[1]
 
         # Format data and send to interpreter
-        resized_image = image.resize((face_input_width, face_input_height))
+        resized_image = image.resize((face_input_width, face_input_height), Image.ANTIALIAS)
         input_data = np.expand_dims(resized_image, axis=0)
         face_interpreter.set_tensor(face_input_details[0]["index"], input_data)
 
@@ -82,13 +77,14 @@ async def predict_face(image: UploadFile = File(...)):
         classes = face_interpreter.get_tensor(face_output_details[1]["index"])[0]
         scores = face_interpreter.get_tensor(face_output_details[2]["index"])[0]
 
+        data = {}
         faces = []
         for i in range(len(scores)):
             if not classes[i] == 0:  # Face
                 continue
             single_face = {}
-            single_face["confidence"] = float(scores[i])
             single_face["userid"] = "unknown"
+            single_face["confidence"] = float(scores[i])
             single_face["y_min"] = int(float(boxes[i][0]) * image_height)
             single_face["x_min"] = int(float(boxes[i][1]) * image_width)
             single_face["y_max"] = int(float(boxes[i][2]) * image_height)
@@ -107,15 +103,14 @@ async def predict_face(image: UploadFile = File(...)):
 
 @app.post(OBJ_DETECTION_URL)
 async def predict_object(image: UploadFile = File(...)):
-    data = {"success": False}
     try:
         contents = await image.read()
-        image = Image.open(io.BytesIO(contents))  # A PIL image
+        image = Image.open(io.BytesIO(contents))
         image_width = image.size[0]
         image_height = image.size[1]
 
         # Format data and send to interpreter
-        resized_image = image.resize((obj_input_width, obj_input_height))
+        resized_image = image.resize((obj_input_width, obj_input_height), Image.ANTIALIAS)
         input_data = np.expand_dims(resized_image, axis=0)
         obj_interpreter.set_tensor(obj_input_details[0]["index"], input_data)
 
@@ -125,6 +120,7 @@ async def predict_object(image: UploadFile = File(...)):
         classes = obj_interpreter.get_tensor(obj_output_details[1]["index"])[0]
         scores = obj_interpreter.get_tensor(obj_output_details[2]["index"])[0]
 
+        data = {}
         objects = []
         for i in range(len(scores)):
             single_object = {}
@@ -149,21 +145,13 @@ async def predict_object(image: UploadFile = File(...)):
 
 @app.post(SCENE_URL)
 async def predict_scene(image: UploadFile = File(...)):
-    data = {"success": False}
     try:
         contents = await image.read()
-        image = Image.open(io.BytesIO(contents))  # A PIL image
-        # Format data and send to interpreter
-        resized_image = image.resize(
-            (scene_input_width, scene_input_height), Image.ANTIALIAS
-        )
+        image = Image.open(io.BytesIO(contents))
+        resized_image = image.resize((scene_input_width, scene_input_height), Image.ANTIALIAS)
         results = classify_image(scene_interpreter, image=resized_image)
-
-        print(
-            f"results[0]: {results[0]}", file=sys.stderr,
-        )
         label_id, prob = results[0]
-
+        data = {}
         data["label"] = scene_labels[label_id]
         data["confidence"] = prob
         data["success"] = True