Roboflow trained Model does not find predections on OAK-D-Lite

I have trained a model using Roboflow work space. The model works fine with web cam web interface.

When I try running the inference on OAK-D-Lite from below code. I can see the stream but no predictions found. I would appreciate any inputs regarding this.

from roboflowoak import RoboflowOak
import cv2
import time
import numpy as np

if name == ‘main’:
# instantiating an object (rf) with the RoboflowOak module
rf = RoboflowOak(model=“airport-checked-bags”, confidence=0.05, overlap=0.5,
version=“1”, api_key=“gO1Aqw06dd77Uq46Jzre”, rgb=True,
depth=True, device=None, blocking=True)
# Running our model and displaying the video output with detections
while True:
t0 = time.time()
# The rf.detect() function runs the model inference
result, frame, raw_frame, depth = rf.detect()
predictions = result[“predictions”]
# predictions:
# [ {
# x: (middle),
# y:(middle),
# width:
# height:
# depth: ###->
# confidence:
# class:
# mask: {
# ]
#frame - frame after preprocs, with predictions
#raw_frame - original frame from your OAK
#depth - depth map for raw_frame, center-rectified to the center camera

    # timing: for benchmarking purposes
    t = time.time()-t0
    print("INFERENCE TIME IN MS ", 1/t)
    print("PREDICTIONS ", [p.json() for p in predictions])

    # setting parameters for depth calculation
    max_depth = np.amax(depth)
    cv2.imshow("depth", depth/max_depth)
    # displaying the video feed as successive frames
    cv2.imshow("frame", frame)

    # how to close the OAK inference window / stop inference: CTRL+q or CTRL+c
    if cv2.waitKey(1) == ord('q'):

Hi @sandeep406

Sorry to hear you’re having issues. I’ve filed a bug report with the team and hopefully, it’ll be fixed soon. In the meantime, you could consider some alternative deployment options.

I actually encounter the same problem. Is this bug still being fixed?

@sandeep406: We noticed you posted your private API key in your post. We’ve revoked the use of that key. You can generate a new one in your workspace settings.

Your API key can be used to run inference, upload images, get info, and almost anything regarding your account and should never be posted publicly. We don’t want your accounts getting accessed by unauthorized people, so please be careful in the future!

Hi @sandeep406 and @KoKonutZzz

We pushed a fix for our OAK deployment last week. Can you confirm that you’re using version 0.0.11 (or later)? You can update using pip:

pip install roboflowoak==0.0.11

Hi @leo
Thank you for the update, but it still failed this time, with a different feedback:

Traceback (most recent call last):
  File "", line 25, in <module>
    rf = RoboflowOak(model="dirt-overflowing", confidence=0.05, overlap=0.5,
  File "/Users/cloudscapespare/anaconda3/envs/csl-load-count-ml/lib/python3.8/site-packages/roboflowoak/", line 59, in __init__
    self.cache_path = self.find_weights()
  File "/Users/cloudscapespare/anaconda3/envs/csl-load-count-ml/lib/python3.8/site-packages/roboflowoak/", line 135, in find_weights
    return download_blob(
  File "/Users/cloudscapespare/anaconda3/envs/csl-load-count-ml/lib/python3.8/site-packages/roboflowoak/", line 25, in download_blob
    raise Exception(str(api_data.json()))
Exception: {'error': {'message': 'This model endpoint is over your the device limit of your workspace.', 'type': 'QuotaException', 'hint': 'If have reached this in error, please let us know.'}}

I tried both using a OAK-D Lite and a OAK-D S2, not sure if other users have had the same problem. Any thoughts please?


This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.