杨珉

I am stuck by training and find that Tensorflow2.0 installed by pip does not support for GPU

How to prepare coco data in tfrecord format?

1

Your code does not contain the path for the original coco detection dataset. And the downloading of tfrecord format coco is wrong:“E tensorflow/tsl/platform/cloud/curl_http_request.cc:610] The transmission of request 0x9690640 (URI: https://www.googleapis.com/storage/v1/b/pix2seq/o?fields=items%2Fname%2CnextPageToken&prefi...

questions about mAP results of anet dataset

2

In the paper, your results is 35.6%. However, you provide three different tools("builtin" "cuhk" "gt"), which one produce results of 35.6%

I found you did not use the self.cls_nonlocal() and self.reg_nonlocal in roi_box_feature_extractors.py.The number of blocks "self.cls_num_stack" and "self.reg_num_stack" are zero

1

codes are here:https://github.com/wuyuebupt/doubleheadsrcnn/blob/master/maskrcnn_benchmark/modeling/roi_heads/box_head/roi_box_feature_extractors.py. And another question:I found no difference between the reg and cls operations. All the network structure parameters are the same except some details.

AttributeError: 'SyncBatchNorm' object has no attribute '_specify_ddp_gpu_num'

6

I got the error explained above.Is this function "_specify_ddp_gpu_num" exist？Thank you!

Do you have nms process during testing the results?

4

I use test.py to get the result of [email protected], but I can not find the nms process which is necessary for evaluating the result.

Low mAP on Activitynet-1.3

8

When I run all the codes to obtain detection_result_nms0.6.json, the detection results is bad. 

Google drive is empty!