Converting UIImage Into CVPixelBuffer and Feeding Into MediaPipe Face Mesh Returns No Result

Question

Problem

In short, I am trying to use the Face Mesh SDK on some PNG images.

Steps

1. Get the image as a UIImage: UIImage(named:"image_name")
2. Convert the UIImage into a CVPixelBuffer using the following extension:

extension UIImage {
   
    func toPixelBuffer() -> CVPixelBuffer? {
        let attrs = [kCVPixelBufferCGImageCompatibilityKey: kCFBooleanTrue, kCVPixelBufferCGBitmapContextCompatibilityKey: kCFBooleanTrue] as CFDictionary
        var pixelBuffer : CVPixelBuffer?
        let status = CVPixelBufferCreate(kCFAllocatorDefault, Int(size.width), Int(size.height), kCVPixelFormatType_32ARGB, attrs, &pixelBuffer)
        guard (status == kCVReturnSuccess) else {
          return nil
        }

        CVPixelBufferLockBaseAddress(pixelBuffer!, CVPixelBufferLockFlags(rawValue: 0))
        let pixelData = CVPixelBufferGetBaseAddress(pixelBuffer!)

        let rgbColorSpace = CGColorSpaceCreateDeviceRGB()
        let context = CGContext(data: pixelData, width: Int(size.width), height: Int(size.height), bitsPerComponent: 8, bytesPerRow: CVPixelBufferGetBytesPerRow(pixelBuffer!), space: rgbColorSpace, bitmapInfo: CGImageAlphaInfo.noneSkipFirst.rawValue)

        context?.translateBy(x: 0, y: size.height)
        context?.scaleBy(x: 1.0, y: -1.0)

        UIGraphicsPushContext(context!)
        draw(in: CGRect(x: 0, y: 0, width: size.width, height: size.height))
        UIGraphicsPopContext()
        CVPixelBufferUnlockBaseAddress(pixelBuffer!, CVPixelBufferLockFlags(rawValue: 0))

        return pixelBuffer
    }
}

3. Setup FaceMesh:

faceMesh = FaceMesh()
faceMesh.startGraph()
faceMesh.delegate = self

4. Feed the pixel buffer (from step 2) into Mediapipe FaceMesh:

if let frame = image.toPixelBuffer() {
    lastFrameBuffer = frame
    faceMesh.processVideoFrame(frame)
}

5. The frame passes into this function:

- (void)processVideoFrame:(CVPixelBufferRef)imageBuffer {
    NSLog(@"[FACEMESH] Processing Video Frame %@", kGraphName);
    const auto ts =
    mediapipe::Timestamp(self.timestamp++ * mediapipe::Timestamp::kTimestampUnitsPerSecond);
    NSError* err = nil;
    BOOL sent = [self.mediapipeGraph sendPixelBuffer:imageBuffer
                                          intoStream:kInputStream
                                          packetType:MPPPacketTypePixelBuffer
                                           timestamp:ts
                                      allowOverwrite:NO
                                               error:&err];
    NSLog(@"[FACEMESH] Sent? %d", sent);
    if (err == nil) {
        NSLog(@"[FACEMESH] NO ERROR");
    } else {
        NSLog(@"[FACEMESH] ERROR: %@",[err localizedDescription]);
    }
}

Result

The logs show that all the steps above have passed

[FACEMESH] inited graph pure_face_mesh_mobile_gpu
[FACEMESH] Started graph pure_face_mesh_mobile_gpu
[FACEMESH] Processing Video Frame pure_face_mesh_mobile_gpu
[FACEMESH] Sent? 1
[FACEMESH] NO ERROR

HOWEVER, and here's the problem, the delegate functions are not being triggered, and the following function is not reached in the c++ code

- (void)mediapipeGraph:(MPPGraph*)graph
       didOutputPacket:(const ::mediapipe::Packet&)packet
            fromStream:(const std::string&)streamName {
    NSLog(@"[FACEMESH] Received Output Packet %@", kGraphName);
    if (streamName == kLandmarksOutputStream) {
        if (packet.IsEmpty()) {
            NSLog(@"[FACEMESH] Packet is empty for the given input %@", kGraphName);
            return;
        }
...

Note that the same process works fine on CVPixelBuffers received from live camera frames.

What could be causing the converted frames to behave this way?

Thanks for any help!