error：：：：AttributeError: 'module' object has no attribute 'make_boxes'

[jiagh2010]

boxes = dn.make_boxes(net)

AttributeError: 'module' object has no attribute 'make_boxes'

[shadowkane]

the whole project doesn't have make_boxes() function, maybe this was in older version of the darknet and the new one doesn't include it but they didn't change the python example. and if you look close you can find a struct type call detection has a attribute of box type you can use it with the 'get_network_boxes' function. detection get_network_boxes(network net, int w, int h, float thresh, float hier, int map, int relative, int num); this one in c (look darknet.h) and it's already defined in darknet.py, so when you import it you can use it. in short words, look at the function detect in darknet.py, you can just use that function ony by passing an image (type image) and then you will get your bounding boxes. if you get an empty table just add this to your darknet.py

srand = lib.srand srand.argtypes = [c_int] nnp_initialize = lib.nnp_initialize

and before you call detect use this srand(2222222) nnp_initialize()

[laissaouibrahim]

Thank you . This solution helps me to resolve the problem of ( NNPACK ERROR!: 50 ) here is my code to run Real-Time object detection :

RealTime_OD_NNPACK.txt

**My Ambition***** : To acheive 7 pfs on a raspberry :)