how can I get the image with the bounding boxes like this

[lhs21]

I can evaluate the results with eval_caltech , and I get the eval-newOcc=heavy.txt/eval-newAll.txt/eval-newReasonable.txt successfully. and the MR-2 are good， 45.7 ， 57.6 ， 4.65 。But when I try to use the results which is in the file(output/valresults/caltech/h/off/) to draw bounding boxes on the test image.I get the image like this . it has too many wrong bounding boxex. I don't know what's the matter . the code I write is added in test_caltech.py.

row=np.shape(boxes)[0]
for i in range(0,row):
....a=boxes[i]
....cv2.rectangle(img,(a[0],a[1]),(a[2],a[3]),(0,255,0), 2)
cv2.imwrite('./test/' + str(set)+str(video)+str(frame_number-1) + '.jpg', img)

i'm writing the graduation thesis ，so it'll help me a lot if you can tell me what's the problem,
thank you very much

[lhs21]

another dataset：citypersons，
how can I get the image with the bounding boxes

[sunshineInmoon]

你好，我也遇到了相同的问题，请问你解决了吗？