RuntimeError: Unable to load the engine file- YOLOV4-Tiny Custom Weights

I am trying to convert custom weights from **yolov4-tiny model** . When I try and build an engine I am getting following error. The YOLO anchors are also setup correctly. Can anyone help me with this ? @GeekAlexis

root@1bf3cb8ffb41:~/FastMOT# python3 app.py --input_uri mid1.mp4 --mot -g
[ WARN:0] global /usr/src/app/OpenCV/modules/videoio/src/cap_gstreamer.cpp (896) open OpenCV | GStreamer warning: unable to query duration of stream
[ WARN:0] global /usr/src/app/OpenCV/modules/videoio/src/cap_gstreamer.cpp (933) open OpenCV | GStreamer warning: Cannot query video position: status=1, value=0, duration=-1
2021-07-15 12:23:48 [    INFO] 1280x720 stream @ 30 FPS
2021-07-15 12:23:48 [    INFO] Loading detector model...
2021-07-15 12:23:49 [    INFO] Building engine with batch size: 1
2021-07-15 12:23:49 [    INFO] This may take a while...
[TensorRT] ERROR: ../rtSafe/safeRuntime.cpp (25) - Cuda Error in allocate: 2 (out of memory)
[TensorRT] ERROR: ../rtSafe/safeRuntime.cpp (25) - Cuda Error in allocate: 2 (out of memory)
2021-07-15 12:23:51 [CRITICAL] Failed to build engine
Traceback (most recent call last):
  File "app.py", line 98, in <module>
    main()
  File "app.py", line 50, in main
    mot = fastmot.MOT(config['resize_to'], config['mot'], draw=draw, verbose=args.verbose)
  File "/usr/src/app/FastMOT/fastmot/mot.py", line 55, in __init__
    self.detector = YOLODetector(self.size, config['yolo_detector'])
  File "/usr/src/app/FastMOT/fastmot/detector.py", line 187, in __init__
    self.backend = InferenceBackend(self.model, 1)
  File "/usr/src/app/FastMOT/fastmot/utils/inference.py", line 44, in __init__
    raise RuntimeError('Unable to load the engine file')
RuntimeError: Unable to load the engine file

YOLOV4-Tiny ANCHORS - yolo.py

class YOLOv4(YOLO):
    ENGINE_PATH = Path(__file__).parent / 'yolov4-tiny.trt' #'yolov4_crowdhuman.trt'
    MODEL_PATH = Path(__file__).parent /  'yolov4-tiny.onnx' #'yolov4_crowdhuman.onnx'

    NUM_CLASSES = 12
    INPUT_SHAPE = (3, 416, 416)
    LAYER_FACTORS = [32, 16]
    SCALES = [1.05, 1.05]
    ANCHORS = [[10, 14, 23, 27, 37, 58],
               [81, 82, 135, 169, 344, 319]]
    #ANCHORS = [[81, 82, 135, 169, 344, 319],
    #           [23, 27, 37, 58, 81, 82]]

Classes
12

yolo.cfg

[net]
# Testing
#batch=1
#subdivisions=1
# Training
batch=64
subdivisions=1
width=416
height=416
channels=3
momentum=0.9
decay=0.0005
angle=0
saturation = 1.5
exposure = 1.5
hue=.1

learning_rate=0.00261
burn_in=1000
max_batches = 24000
policy=steps
steps=19200,21600
scales=.1,.1

[convolutional]
batch_normalize=1
filters=32
size=3
stride=2
pad=1
activation=leaky

[convolutional]
batch_normalize=1
filters=64
size=3
stride=2
pad=1
activation=leaky

[convolutional]
batch_normalize=1
filters=64
size=3
stride=1
pad=1
activation=leaky

[route]
layers=-1
groups=2
group_id=1

[convolutional]
batch_normalize=1
filters=32
size=3
stride=1
pad=1
activation=leaky

[convolutional]
batch_normalize=1
filters=32
size=3
stride=1
pad=1
activation=leaky

[route]
layers = -1,-2

[convolutional]
batch_normalize=1
filters=64
size=1
stride=1
pad=1
activation=leaky

[route]
layers = -6,-1

[maxpool]
size=2
stride=2

[convolutional]
batch_normalize=1
filters=128
size=3
stride=1
pad=1
activation=leaky

[route]
layers=-1
groups=2
group_id=1

[convolutional]
batch_normalize=1
filters=64
size=3
stride=1
pad=1
activation=leaky

[convolutional]
batch_normalize=1
filters=64
size=3
stride=1
pad=1
activation=leaky

[route]
layers = -1,-2

[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky

[route]
layers = -6,-1

[maxpool]
size=2
stride=2

[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky

[route]
layers=-1
groups=2
group_id=1

[convolutional]
batch_normalize=1
filters=128
size=3
stride=1
pad=1
activation=leaky

[convolutional]
batch_normalize=1
filters=128
size=3
stride=1
pad=1
activation=leaky

[route]
layers = -1,-2

[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky

[route]
layers = -6,-1

[maxpool]
size=2
stride=2

[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky

##################################

[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky

[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky

[convolutional]
size=1
stride=1
pad=1
filters=51
activation=linear



[yolo]
mask = 3,4,5
anchors = 10,14,  23,27,  37,58,  81,82,  135,169,  344,319
classes=12
num=6
jitter=.3
scale_x_y = 1.05
cls_normalizer=1.0
iou_normalizer=0.07
iou_loss=ciou
ignore_thresh = .7
truth_thresh = 1
random=0
resize=1.5
nms_kind=greedynms
beta_nms=0.6

[route]
layers = -4

[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky

[upsample]
stride=2

[route]
layers = -1, 23

[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky

[convolutional]
size=1
stride=1
pad=1
filters=51
activation=linear

[yolo]
mask = 0,1,2
anchors = 10,14,  23,27,  37,58,  81,82,  135,169,  344,319
classes=12
num=6
jitter=.3
scale_x_y = 1.05
cls_normalizer=1.0
iou_normalizer=0.07
iou_loss=ciou
ignore_thresh = .7
truth_thresh = 1
random=0
resize=1.5
nms_kind=greedynms
beta_nms=0.6

[GeekAlexis]

It looks like you changed the default anchor mask from darknet, so they should be as follows:

    ANCHORS = [[81,82, 135,169, 344,319],
               [10,14, 23,27, 37,58]]

[Myron1996]

Yes I did check the output of this model in darknet and it’s working very well.

https://user-images.githubusercontent.com/72124771/126612486-bbf3189c-a4ff-4a3c-8230-820e54fac076.mp4

I used tensorrt_demos to convert it into ONNX .