Mmdetection: TypeError: 'numpy.float64' object cannot be interpreted as an integer

Created on 23 Mar 2020 · 4Comments · Source: open-mmlab/mmdetection

GETTINGSTRTED.md says I can evaluate the model with pretrained weights by

python tools/test.py configs/pascal_voc/faster_rcnn_r50_fpn_1x_voc.py \
    checkpoints/SOME_CHECKPOINT.pth \
    --eval mAP

but when I did so, the output is
……
Evaluating bbox...
Loading and preparing results...
DONE (t=1.98s)
creating index...
index created!
Traceback (most recent call last):
File "/home/whh/anaconda3/envs/open-mmlab/lib/python3.7/site-packages/numpy/core/function_base.py", line 117, in linspace
num = operator.index(num)
TypeError: 'numpy.float64' object cannot be interpreted as an integer

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "tools/test.py", line 175, in
main()
File "tools/test.py", line 171, in main
dataset.evaluate(outputs, args.eval, **kwargs)
File "/home/whh/projects/mmdetection/mmdet/datasets/coco.py", line 359, in evaluate
cocoEval = COCOeval(cocoGt, cocoDt, iou_type)
File "/home/whh/anaconda3/envs/open-mmlab/lib/python3.7/site-packages/pycocotools/cocoeval.py", line 76, in __init__
self.params = Params(iouType=iouType) # parameters
File "/home/whh/anaconda3/envs/open-mmlab/lib/python3.7/site-packages/pycocotools/cocoeval.py", line 521, in __init__
self.setDetParams()
File "/home/whh/anaconda3/envs/open-mmlab/lib/python3.7/site-packages/pycocotools/cocoeval.py", line 501, in setDetParams
self.iouThrs = np.linspace(.5, 0.95, np.round((0.95 - .5) / .05) + 1, endpoint=True)
File "<__array_function__ internals>", line 6, in linspace
File "/home/whh/anaconda3/envs/open-mmlab/lib/python3.7/site-packages/numpy/core/function_base.py", line 121, in linspace
.format(type(num)))
TypeError: object of type cannot be safely interpreted as an integer.

so what should I do

Source

Whileherham

Most helpful comment

Hi @Whileherham ,
How did you install the pycocotools?
You first can try to reinstall it by the following command and then try to evaluate your model again.
pip install "git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI"

Thank you. I installed pycocotools by downloading the code and 'make install' under PythonAPI since the command in your install.md cannot work in my computer.
I finally solved the problem by modifying the source code 'cocoeval.py'. I replaced the
self.iouThrs = np.linspace(.5, 0.95, np.round((0.95 - .5) / .05) + 1, endpoint=True)
self.recThrs = np.linspace(.0, 1.00, np.round((1.00 - .0) / .01) + 1, endpoint=True)
with
self.iouThrs = np.linspace(.5, 0.95, 10, endpoint=True)
self.recThrs = np.linspace(.0, 1.00, 101, endpoint=True)
and it worked.

Some people say it is because the version of my numpy is too high, and the same problem has occured in other repo and is solved by lower version numpy. So I guess maybe we can restrict the version of numpy in install.md

Whileherham on 24 Mar 2020

👍5

All 4 comments

Hi @Whileherham ,
How did you install the pycocotools?
You first can try to reinstall it by the following command and then try to evaluate your model again.

pip install "git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI"

ZwwWayne on 23 Mar 2020

Hi @Whileherham ,
How did you install the pycocotools?
You first can try to reinstall it by the following command and then try to evaluate your model again.
pip install "git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI"