TI AM62A BYOM Unhandled promise rejection

Hi,

I have imported a Tensor Flow Lite model with BYOM feature. RGB image, not normalized, 26 output labels.

Everything went well at EI, but then I deployed to TI AM62A and Edge Impulse Runner returns:

(node:2053) UnhandledPromiseRejectionWarning: Process exited with null
UnhandledPromiseRejectionWarning: Process exited with null
at emitUnhandledRejectionWarning (internal/process/promises.js:151:15)
at processPromiseRejections (internal/process/promises.js:211:11)
at processTicksAndRejections (internal/process/task_queues.js:98:32)
(node:2053) UnhandledPromiseRejectionWarning: Unhandled promise rejection. This error originated either by throwing inside of an async function without a catch block, or by rejecting a promise which was not handled with .catch(). To termina te the node process on unhandled promise rejection, use the CLI flag --unhandle d-rejections=strict (see Command-line API | Node.js v21.1.0 Documentation ns_mode). (rejection id: 1)

With 10 classes/labels, this error does not appear. Any clues? Thanks

Hi @Bandini,

We need some more information to help you troubleshoot this error:

  • What is your project ID in Edge Impulse Studio?
  • How did you deploy your model? Did you just run edge-impulse-linux-runner on the board?

Hi @shawn_edgeimpulse

Project ID 303898
Deployed via edge-impulse-linux-runner to TI AM62A

With Project ID 270046 (10 labels subset instead of 26) works ok

Thanks

Hi @Bandini,

May I ask, which PSDK version are you using? We don’t support > 9.0 yet in Studio.

// Raul

Hi @Bandini

The max label is by default set to 25 in the SDK. You can build manually instead, by:

  1. Clone our example standalone inferencing linux repo.
  2. Download and unzip the TIDL-RT Library (AM62A) from the Studio and place in the top-level directory of the repo (1)
  3. increase EI_CLASSIFIER_MAX_LABELS_COUNT in edge-impulse-sdk/classifier/ei_classifieri_types.h to a number higher than 25.
  4. Build for AM62A
$ APP_EIM=1 TARGET_AM62A=1 make -j

Thanks Raul! I will do that.