search

rbgirshick / py-faster-rcnn

Faster R-CNN (Python implementation) -- see https://github.com/ShaoqingRen/faster_rcnn for the official MATLAB version
Other
8.13k stars 4.11k forks source link

my test result on voc is really very poor,i do not know why? #514

Open boyflytobeman opened 7 years ago

boyflytobeman commented 7 years ago

i retrain the model based on zf net ,i get very poor result ,can someone tell me why?

image

boyflytobeman commented 7 years ago

the iterations is 10000,i am realy very fused!

boyflytobeman commented 7 years ago

train_net: "models/pascal_voc/ZF/faster_rcnn_end2end/train.prototxt"

base_lr: 0.001 lr_policy: "step" gamma: 0.1 stepsize: 50000 display: 20 average_loss: 100 momentum: 0.9 weight_decay: 0.0005

base_lr: 0.001

lr_policy: "exp"

gamma: 0.999539589 # (0.00001/0.001)^(1/10000)

display: 1

average_loss: 100

momentum: 0.9

weight_decay: 0.0005

We disable standard caffe solver snapshotting and implement our own snapshot

function

snapshot: 1000

We still use the snapshot prefix, though

snapshot_prefix: "zf_faster_rcnn" iter_size: 2

Jennyidream0318 commented 7 years ago

have u already solved this problem, cause i also have this trouble with my experiment. If u have already know the reasons which lead to this result, please give me some help, THX!

boyflytobeman commented 7 years ago

yeah，i have resolve it ，use the model '/py-faster-rcnn/output/faster_rcnn_end2end/voc_2007_trainval' will ok