When I execute a loaded network in my Python program, it crashes with the error "Exception has occurred: KeyError 'boxes'" (I get the same error in my WSL Ubuntu and Raspberry Pi 4). I am implementing the code in this page "https://docs.openvino.ai/latest/notebooks/004-hello-detection-with-output.html#" and, my program reads as follows:
import matplotlib.pyplot as plt
import numpy as np
from openvino.inference_engine import IECore
...
openvino_xml = openvino_model_path + 'person-vehicle-bike-detection-crossroad-1016.xml'
openvino_weights = openvino_model_path + 'person-vehicle-bike-detection-crossroad-1016.bin'
device_name = 'MYRIAD' # try this on the Rpi4
# device_name = 'CPU' # try this on WSL Ubuntu runnin in my laptop
picture = picture_captured_path + 'cars-expressway1.jpg'
ie = IECore()
net = ie.read_network(
model=openvino_xml,
weights=openvino_weights,
)
exec_net = ie.load_network(net, device_name)
input_layer_ir = next(iter(exec_net.input_info))
output_layer_ir = next(iter(exec_net.outputs))
# get input and output names of nodes
input_key = list(exec_net.input_info)[0]
output_key = list(exec_net.outputs.keys())[0]
def printOpenVino_InputOutputItems():
print("Inputs:")
for name, info in exec_net.input_info.items():
print("\tname: {}".format(name))
print("\tshape: {}".format(info.tensor_desc.dims))
# print("\tlayout: {}".format(info.layout))
print("\tprecision: {}\n".format(info.precision))
print("Outputs:")
for name, info in exec_net.outputs.items():
print("\tname: {}".format(name))
print("\tshape: {}".format(info.shape))
print("\tlayout: {}".format(info.layout))
print("\tprecision: {}\n".format(info.precision))
print("input_key: " + input_key)
print("output_key: " + output_key)
# Check values in Inputs and Outputs
printOpenVino_InputOutputItems()
imageFrame = cv.imread(picture)
if (type(imageFrame) == type(None)):
break
# Resize with OpenCV your image if needed to match with net input shape
N, C, H, W = exec_net.input_info[input_layer_ir].tensor_desc.dims
resized_image = cv.resize(src=imageFrame, dsize=(W, H))
# Converting image to NCHW format with FP32 type
input_data = np.expand_dims(resized_image.transpose(2, 0, 1), 0)
# DO INFERENCE
result = exec_net.infer(inputs={input_layer_ir: input_data})
# Extract list of boxes from results
boxes = result["boxes"]
When I run this program, it first executes the printOpenVino_InputOutputItems() method printing these results:
Inputs:
name: input.1
shape: [1, 3, 512, 512]
precision: FP32
Outputs:
name: 653
shape: [1, 1, 200, 7]
layout: NCHW
precision: FP32
input_key: input.1
output_key: 653
Backend QtAgg is interactive backend. Turning interactive mode on.
But then in line 62 of my program (where I try to extract the boxes from the results variable), I get the following error:
Exception has occurred: KeyError
'boxes'
File "/home/winlinuxuser/projects/classify-vehicle/openvino-pi4-vehicle-person.py", line 62
boxes = result["boxes"]
Putting a breakpoint in line 59 (using VSCode), I could read the content in the "result" variable and, it was the following one:
Questions:
- Why am I getting this error?
- How can I extract the boxes from the result variable?
Thank you in advance for all your support and will be waiting for your reply,
Link Copied
Hi RGVGreatCoder,
I noticed that the horizontal-text-detection-0001 contains boxes as one of the Results.
while person-vehicle-bike-detection-crossroad-1016 only contains detection_out as a Result.
You may check it by using Netron.
As a workaround, I suggest you use the Object Detection SSD Python Sample. The demo is located under the following directory:
<INSTALL_DIR>\inference_engine\samples\python\object_detection_sample_ssd
On another note, I have validated the demo application with person-vehicle-bike-detection-crossroad-1016 as shown as follows.
Regards,
Wan
