mxnet
2018.05.17 11:21
senet 18.68% 4.47%
se-50 (SE-ResNeXt-50 (32 x 4d)) 20.97% 5.54%
resnext-101 (64x4d) 20.4% 5.3%
resnet-152 23% 6.7%
1070, 8GB
resnet -> batch 26 , 50img/sec
se-50 -> batch 39 , 35img/sec
resnext -> batch 16 , 25img/sec
댓글 2
번호 | 제목 | 글쓴이 | 날짜 | 조회 수 |
---|---|---|---|---|
1549 | ambient-light-sensor | WHRIA | 2018.09.29 | 280 |
1548 | register | WHRIA | 2018.09.21 | 1931 |
1547 | Deep learning 기반 DEMO | WHRIA | 2018.09.16 | 7121 |
1546 | slide note 삭제 슬라이드 노트 삭제 | WHRIA | 2018.09.12 | 4436 |
1545 | faster rcnn resnet 101 | WHRIA | 2018.09.09 | 175 |
1544 | 나이 40에... | WHRIA | 2018.09.01 | 143 |
1543 | letencrypt - win | WHRIA | 2018.08.12 | 48 |
1542 |
달러 환율
![]() | WHRIA | 2018.06.16 | 1300 |
1541 | Interpretation of the Outputs of Deep Learning Model Trained with Skin Cancer Dataset [1] | WHRIA | 2018.06.02 | 7710 |
1540 | skin cancer [1] | WHRIA | 2018.05.23 | 1472 |
1539 | deep learning framework | WHRIA | 2018.05.17 | 1789 |
» | mxnet [2] | WHRIA | 2018.05.17 | 3214 |
1537 | intel 내장 그래픽 / CUDA | WHRIA | 2018.05.12 | 3159 |
1536 | 당좌대월이자율 | WHRIA | 2018.03.06 | 2602 |
1535 | security 0 | WHRIA | 2018.03.02 | 2473 |
resnet은 32 , se-50 은 56, resnext 는 26 을 batch 로 줄 수 있고 성능은 2배정도 - 1080ti