Why isn't my 1-Dimensional barcode being detected?

First, check that the barcode is of a type supported by AVMetadataMachineReadableCodeObject.

Check that the barcode is in focus. Implementing tap-to-focus or repeatedly triggering auto-focus is recommended for best results on auto-focus enabled cameras. When using a fixed focus camera some codes might be too small to decode at the appropriate focus distance.

Check that the AVCaptureSessionPreset or AVCaptureDeviceFormat provides sufficient resolution for the density of barcode you're trying to scan.

Why is my 1-Dimensional barcode not detected at all locations in the image?

2-dimensional barcode symbologies (Aztec, QR) have easily found marker patterns which allow them to be found efficiently anywhere in a digital image. 1-dimensional barcode symbologies (UPCE, Code39, Code93, Code128, EAN8, EAN13, PDF417*) have no such patterns. This is because they were designed to be read by a laser scanner.

The AVCaptureMetadataOutput 1-dimensional barcode detectors behave in a manner consistent with this design. The detector samples the image using a limited set of scan lines. The AVCaptureMetadataOutput API can detect a barcode only when a decodable portion of the barcode intersects one of these lines.

The scanning pattern includes horizontal, vertical, near-horizontal, and near-vertical scan lines around the center of the AVCaptureMetadataOutput rectOfInterest ("Center"). See Table 1. You may wish to provide hints to the user about where to place the barcode within the camera preview for best results.

In some circumstances, when fewer symbologies are enabled and on more powerful hardware, the set of scan lines is expanded to cover a larger portion of the region of interest ("Additional").

* PDF417 is a 2-dimensional symbology, but is treated by AVCaptureMetadataOutput as a (stacked) 1-dimensional code.

Can AVCaptureMetadataOutput detect multiple 1-Dimensional barcodes in a frame?

No. Only a single 1-dimensional code will be returned with each call to AVCaptureMetadataOutputObjectDelegate captureOutput:didOutputMetadataObjects:fromConnection:, and it will correspond to the center-most decodable barcode in the rectOfInterest.

Can AVCaptureMetadataOutput detect multiple 2-Dimensional barcodes in a frame?

Yes. There is a limit of four barcodes per call to AVCaptureMetadataOutputObjectDelegate captureOutput:didOutputMetadataObjects:fromConnection: which will usually correspond to the center-most codes in the rectOfInterest.

Is UPC-A supported?

Yes. UPC-A is a formal subset of EAN-13. A decoded UPC-A barcode is output as AVMetadataObjectTypeEAN13Code with a leading zero in the stringValue.

Are EAN-2 and EAN-5 supported?

No.

How can I improve scanning results on very small or very dense codes?

Code detectability depends on many factors, including: capture resolution, amount of light, complexity / density of the code being scanned, and distance from the lens.

1. Try using a higher resolution preset, but avoid using AVCaptureSessionPresetPhoto resolution — this can lead to longer exposure times and increased blur in low-light situations.

2. Try setting a videoZoomFactor between 1.0 and the videoZoomFactorUpscaleThreshold. This provides a higher resolution window on a smaller part of the image, which means there are more barcode pixels for the decoder to work with.

You should experiment with different camera resolutions and code densities to find the right combination that works for your particular application.