Merge pull request #493 from luxonis/depth_docs

Erol444 · web-flow · commit 9623e6ad8f74 · 2022-01-27T20:23:40.000+01:00
Added some clarifications to depth docs
diff --git a/depthai-core b/depthai-core
@@ -1 +1 @@
-Subproject commit 0caf71c904606fb34116c66cc8e999e0ee17d863
+Subproject commit ea769db29428229c496fc1b3dd61161858671174
diff --git a/docs/source/components/nodes/image_manip.rst b/docs/source/components/nodes/image_manip.rst
@@ -72,7 +72,7 @@ ImageManip node supports the following image formats:
 
 - for RGB input/output types only :code:`RGB888p` / :code:`BGR888p` (planar, not interleaved)
 - :code:`RAW16` input, but can only for cropping (use-case: depth map), outputting same format
-- can output :code:`GRAY8` (either as single plane, or 3-plane RGB as set with setFrameType()), but only from these inputs: :code:`RAW8, :code:`GRAY8`, :code:`YUV420p` (:ref:`ColorCamera` isp), :code:`NV12` (:ref:`ColorCamera` video / still)
+- can output :code:`GRAY8` (either as single plane, or 3-plane RGB as set with setFrameType()), but only from these inputs: :code:`RAW8`, :code:`GRAY8`, :code:`YUV420p` (:ref:`ColorCamera` isp), :code:`NV12` (:ref:`ColorCamera` video / still)
 
 Limitations
 ###########
diff --git a/docs/source/components/nodes/neural_network.rst b/docs/source/components/nodes/neural_network.rst
@@ -117,6 +117,8 @@ Usage
 Examples of functionality
 #########################
 
+- :ref:`Multi-Input Frame Concationation`
+- :ref:`Frame Normalization`
 - `DeeplabV3 experiment <https://github.com/luxonis/depthai-experiments/tree/master/gen2-deeplabv3_depth>`__
 - `Age/gender experiment <https://github.com/luxonis/depthai-experiments/blob/master/gen2-age-gender/main.py>`__
 - `EfficientDet demo <https://github.com/luxonis/depthai-experiments/blob/master/gen2-efficientDet/main.py>`__
diff --git a/docs/source/components/nodes/stereo_depth.rst b/docs/source/components/nodes/stereo_depth.rst
@@ -31,7 +31,7 @@ Inputs and Outputs
                  │                   ├─────────────►
   left           │                   │   syncedLeft
   ──────────────►│-------------------├─────────────►
-                 │                   │        depth
+                 │                   │   depth [mm]
                  │                   ├─────────────►
                  │    StereoDepth    │    disparity
                  │                   ├─────────────►
@@ -56,8 +56,8 @@ Inputs and Outputs
     - :code:`confidenceMap` - :ref:`ImgFrame`
     - :code:`rectifiedLeft` - :ref:`ImgFrame`
     - :code:`syncedLeft` - :ref:`ImgFrame`
-    - :code:`depth` - :ref:`ImgFrame`
-    - :code:`disparity` - :ref:`ImgFrame`
+    - :code:`depth` - :ref:`ImgFrame`: UINT16 values - depth in millimeters
+    - :code:`disparity` - :ref:`ImgFrame`: UINT8 or UINT16 if Subpixel mode
     - :code:`rectifiedRight` - :ref:`ImgFrame`
     - :code:`syncedRight` - :ref:`ImgFrame`
     - :code:`outConfig` - :ref:`StereoDepthConfig`
@@ -233,6 +233,7 @@ Examples of functionality
 #########################
 
 - :ref:`Depth Preview`
+- :ref:`RGB Depth alignment`
 - :ref:`Mono & MobilenetSSD & Depth`
 - :ref:`RGB & MobilenetSSD with spatial data`
 
diff --git a/docs/source/samples/StereoDepth/rgb_depth_aligned.rst b/docs/source/samples/StereoDepth/rgb_depth_aligned.rst
@@ -1,13 +1,13 @@
-RGB Depth
-=========
+RGB Depth alignment
+===================
 
-This example shows usage of RGB depth. Since our boards has 1 color and 2 mono cameres, therefore
-you need 2 mono cameras for the depth map, then you need to align the image from the
-color camera on top of that to get RGB depth.
+This example shows usage of RGB depth alignment. Since OAK-D has a color and a pair of stereo cameras,
+you can align depth map to the color frame on top of that to get RGB depth.
 
 Demo
 ####
 
+.. image:: https://user-images.githubusercontent.com/18037362/151351377-a5752fbe-3b8b-4985-b8d1-d5f8a7d5a868.png
 
 Setup
 #####
diff --git a/docs/source/tutorials/code_samples.rst b/docs/source/tutorials/code_samples.rst
@@ -132,7 +132,7 @@ are presented with code.
 - :ref:`Depth Preview` - Displays colorized stereo disparity
 - :ref:`Stereo Depth from host` - Generates stereo depth frame from a set of mono images from the host
 - :ref:`Stereo Depth Video` - An extended version of **Depth Preview**
-- :ref:`RGB Depth` - Displays RGB depth frames
+- :ref:`RGB Depth alignment` - Displays RGB depth aligned frames
 
 .. rubric:: SystemLogger
 
diff --git a/examples/SpatialDetection/spatial_location_calculator.py b/examples/SpatialDetection/spatial_location_calculator.py
@@ -74,7 +74,8 @@
     while True:
         inDepth = depthQueue.get() # Blocking call, will wait until a new data has arrived
 
-        depthFrame = inDepth.getFrame()
+        depthFrame = inDepth.getFrame() # depthFrame values are in millimeters
+
         depthFrameColor = cv2.normalize(depthFrame, None, 255, 0, cv2.NORM_INF, cv2.CV_8UC1)
         depthFrameColor = cv2.equalizeHist(depthFrameColor)
         depthFrameColor = cv2.applyColorMap(depthFrameColor, cv2.COLORMAP_HOT)
diff --git a/examples/SpatialDetection/spatial_mobilenet.py b/examples/SpatialDetection/spatial_mobilenet.py
@@ -110,7 +110,8 @@
             startTime = current_time
 
         frame = inPreview.getCvFrame()
-        depthFrame = depth.getFrame()
+
+        depthFrame = depth.getFrame() # depthFrame values are in millimeters
 
         depthFrameColor = cv2.normalize(depthFrame, None, 255, 0, cv2.NORM_INF, cv2.CV_8UC1)
         depthFrameColor = cv2.equalizeHist(depthFrameColor)
diff --git a/examples/SpatialDetection/spatial_mobilenet_mono.py b/examples/SpatialDetection/spatial_mobilenet_mono.py
@@ -118,7 +118,8 @@
 
         rectifiedRight = inRectified.getCvFrame()
 
-        depthFrame = inDepth.getFrame()
+        depthFrame = inDepth.getFrame() # depthFrame values are in millimeters
+
         depthFrameColor = cv2.normalize(depthFrame, None, 255, 0, cv2.NORM_INF, cv2.CV_8UC1)
         depthFrameColor = cv2.equalizeHist(depthFrameColor)
         depthFrameColor = cv2.applyColorMap(depthFrameColor, cv2.COLORMAP_HOT)
diff --git a/examples/SpatialDetection/spatial_tiny_yolo.py b/examples/SpatialDetection/spatial_tiny_yolo.py
@@ -132,7 +132,8 @@
         depth = depthQueue.get()
 
         frame = inPreview.getCvFrame()
-        depthFrame = depth.getFrame()
+        depthFrame = depth.getFrame() # depthFrame values are in millimeters
+
         depthFrameColor = cv2.normalize(depthFrame, None, 255, 0, cv2.NORM_INF, cv2.CV_8UC1)
         depthFrameColor = cv2.equalizeHist(depthFrameColor)
         depthFrameColor = cv2.applyColorMap(depthFrameColor, cv2.COLORMAP_HOT)
diff --git a/examples/StereoDepth/depth_crop_control.py b/examples/StereoDepth/depth_crop_control.py
@@ -59,7 +59,8 @@
 
     while True:
         inDepth = q.get()
-        depthFrame = inDepth.getFrame()
+        depthFrame = inDepth.getFrame() # depthFrame values are in millimeters
+
         # Frame is transformed, the color map will be applied to highlight the depth info
         depthFrameColor = cv2.normalize(depthFrame, None, 255, 0, cv2.NORM_INF, cv2.CV_8UC1)
         depthFrameColor = cv2.equalizeHist(depthFrameColor)
diff --git a/examples/StereoDepth/depth_post_processing.py b/examples/StereoDepth/depth_post_processing.py
@@ -60,8 +60,8 @@
     q = device.getOutputQueue(name="disparity", maxSize=4, blocking=False)
 
     while True:
-        inDepth = q.get()  # blocking call, will wait until a new data has arrived
-        frame = inDepth.getFrame()
+        inDisparity = q.get()  # blocking call, will wait until a new data has arrived
+        frame = inDisparity.getFrame()
         # Normalization for better visualization
         frame = (frame * (255 / depth.initialConfig.getMaxDisparity())).astype(np.uint8)
 
diff --git a/examples/StereoDepth/depth_preview.py b/examples/StereoDepth/depth_preview.py
@@ -48,8 +48,8 @@
     q = device.getOutputQueue(name="disparity", maxSize=4, blocking=False)
 
     while True:
-        inDepth = q.get()  # blocking call, will wait until a new data has arrived
-        frame = inDepth.getFrame()
+        inDisparity = q.get()  # blocking call, will wait until a new data has arrived
+        frame = inDisparity.getFrame()
         # Normalization for better visualization
         frame = (frame * (255 / depth.initialConfig.getMaxDisparity())).astype(np.uint8)
 
diff --git a/examples/StereoDepth/rgb_depth_aligned.py b/examples/StereoDepth/rgb_depth_aligned.py
@@ -5,8 +5,8 @@
 import depthai as dai
 
 # Weights to use when blending depth/rgb image (should equal 1.0)
-rgbWeight = 0.6
-depthWeight = 0.4
+rgbWeight = 0.4
+depthWeight = 0.6
 
 
 def updateBlendWeights(percent_rgb):
@@ -39,12 +39,12 @@ def updateBlendWeights(percent_rgb):
 stereo = pipeline.create(dai.node.StereoDepth)
 
 rgbOut = pipeline.create(dai.node.XLinkOut)
-depthOut = pipeline.create(dai.node.XLinkOut)
+disparityOut = pipeline.create(dai.node.XLinkOut)
 
 rgbOut.setStreamName("rgb")
 queueNames.append("rgb")
-depthOut.setStreamName("depth")
-queueNames.append("depth")
+disparityOut.setStreamName("disp")
+queueNames.append("disp")
 
 #Properties
 camRgb.setBoardSocket(dai.CameraBoardSocket.RGB)
@@ -71,16 +71,13 @@ def updateBlendWeights(percent_rgb):
 camRgb.isp.link(rgbOut.input)
 left.out.link(stereo.left)
 right.out.link(stereo.right)
-stereo.disparity.link(depthOut.input)
+stereo.disparity.link(disparityOut.input)
 
 # Connect to device and start pipeline
 with dai.Device(pipeline) as device:
 
-    device.getOutputQueue(name="rgb",   maxSize=4, blocking=False)
-    device.getOutputQueue(name="depth", maxSize=4, blocking=False)
-
     frameRgb = None
-    frameDepth = None
+    frameDisp = None
 
     # Configure windows; trackbar adjusts blending ratio of rgb/depth
     rgbWindowName = "rgb"
@@ -94,9 +91,9 @@ def updateBlendWeights(percent_rgb):
     while True:
         latestPacket = {}
         latestPacket["rgb"] = None
-        latestPacket["depth"] = None
+        latestPacket["disp"] = None
 
-        queueEvents = device.getQueueEvents(("rgb", "depth"))
+        queueEvents = device.getQueueEvents(("rgb", "disp"))
         for queueName in queueEvents:
             packets = device.getOutputQueue(queueName).tryGetAll()
             if len(packets) > 0:
@@ -106,25 +103,25 @@ def updateBlendWeights(percent_rgb):
             frameRgb = latestPacket["rgb"].getCvFrame()
             cv2.imshow(rgbWindowName, frameRgb)
 
-        if latestPacket["depth"] is not None:
-            frameDepth = latestPacket["depth"].getFrame()
+        if latestPacket["disp"] is not None:
+            frameDisp = latestPacket["disp"].getFrame()
             maxDisparity = stereo.initialConfig.getMaxDisparity()
             # Optional, extend range 0..95 -> 0..255, for a better visualisation
-            if 1: frameDepth = (frameDepth * 255. / maxDisparity).astype(np.uint8)
+            if 1: frameDisp = (frameDisp * 255. / maxDisparity).astype(np.uint8)
             # Optional, apply false colorization
-            if 1: frameDepth = cv2.applyColorMap(frameDepth, cv2.COLORMAP_HOT)
-            frameDepth = np.ascontiguousarray(frameDepth)
-            cv2.imshow(depthWindowName, frameDepth)
+            if 1: frameDisp = cv2.applyColorMap(frameDisp, cv2.COLORMAP_HOT)
+            frameDisp = np.ascontiguousarray(frameDisp)
+            cv2.imshow(depthWindowName, frameDisp)
 
         # Blend when both received
-        if frameRgb is not None and frameDepth is not None:
+        if frameRgb is not None and frameDisp is not None:
             # Need to have both frames in BGR format before blending
-            if len(frameDepth.shape) < 3:
-                frameDepth = cv2.cvtColor(frameDepth, cv2.COLOR_GRAY2BGR)
-            blended = cv2.addWeighted(frameRgb, rgbWeight, frameDepth, depthWeight, 0)
+            if len(frameDisp.shape) < 3:
+                frameDisp = cv2.cvtColor(frameDisp, cv2.COLOR_GRAY2BGR)
+            blended = cv2.addWeighted(frameRgb, rgbWeight, frameDisp, depthWeight, 0)
             cv2.imshow(blendedWindowName, blended)
             frameRgb = None
-            frameDepth = None
+            frameDisp = None
 
         if cv2.waitKey(1) == ord('q'):
             break
