reset

Create a RAFT optical flow object.

flowModel = opticalFlowRAFT;

Create an object to read the input video file.

vidReader = VideoReader("visiontraffic.avi",CurrentTime=11);

Create a custom figure window to visualize the optical flow vectors.

h = figure;
movegui(h);
hViewPanel = uipanel(h, Position=[0 0 1 1], Title="Plot of Optical Flow Vectors");
hPlot = axes(hViewPanel);

Read consecutive image frames to estimate optical flow. Display the current frame and overlay optical flow vectors using a quiver plot. The estimateFlow function calculates the optical flow between two consecutive frames.

Note that the function internally stores the previous frame and utilizes it implicitly for optical flow estimation. Consequently, when the function is called for the first time on a sequence of frames, it will return a zero flow. This is because, in the absence of a genuine previous frame, the initial frame is treated as both the current and previous frame, leading to no detectable motion between the two. This is consistent with the argument structure and behavior of established optical flow estimation methods, such as opticalFlowFarneback.

while hasFrame(vidReader)
    frame = readFrame(vidReader);
    flow = estimateFlow(flowModel,frame);

    imshow(frame)
    hold on
    plot(flow,DecimationFactor=[10 10],ScaleFactor=0.45,Parent=hPlot,color="g");
    hold off
    pause(10^-3)
end

Reset the opticalFlowRAFT object after the video processing has completed. This clears the internal state of the object, including the saved previous frame.

reset(flowModel);

RAFT provides accurate optical flow estimation, but it can be memory-intensive when applied to large images or run on GPUs with limited memory. In such cases, processing high-resolution frames directly may lead to out-of-memory (OOM) errors. A practical solution is to downscale the input images before passing them to the RAFT model, compute optical flow at the reduced resolution, and then rescale the resulting flow field back to the original size of the imagery. This example demonstrates how to apply this memory-efficient strategy, allowing you to use RAFT on larger images or resource-constrained hardware while preserving meaningful flow estimates.

I1 = imread("yellowstone_left.png");
I2 = imread("yellowstone_right.png");

figure
imshow(stereoAnaglyph(I1, I2))
title("Composite Image (Red - Left Image, Cyan - Right Image)")

disp(size(I1, [1 2]))

   480   640

raft = opticalFlowRAFT;

estimateFlow(raft, I1);
flowFull = estimateFlow(raft, I2);

I1Small = imresize(I1, 0.5);
I2Small = imresize(I2, 0.5);

disp(size(I1Small, [1 2]))

   240   320

reset(raft);
estimateFlow(raft, I1Small);
flowLowRes = estimateFlow(raft, I2Small);

origDims = size(I1, [1 2]);
scaleX = origDims(2) / size(flowLowRes.Magnitude, 2);
scaleY = origDims(1) / size(flowLowRes.Magnitude, 1);

flowX = imresize(flowLowRes.Vx, origDims, "bilinear") * scaleX;
flowY = imresize(flowLowRes.Vy, origDims, "bilinear") * scaleY;

flowUpscaled = opticalFlow(flowX, flowY);

figure
subplot(1,2,1)
imshow(I1)
hold on
plot(flowFull, DecimationFactor=[40 40], ScaleFactor=0.75, Color="g");
hold off
title("Optical Flow at Full Resolution")

subplot(1,2,2)
imshow(I1)
hold on
plot(flowUpscaled, DecimationFactor=[40 40], ScaleFactor=0.75, Color="g");
hold off
title("Optical Flow at Half Resolution")