search

thtrieu / darkflow

Translate darknet to tensorflow. Load trained weights, retrain/fine-tune using tensorflow, export constant graph def to mobile devices
GNU General Public License v3.0
6.13k stars 2.08k forks source link

No results whatsoever #1022

Open Sequential-circuits opened 5 years ago

Sequential-circuits commented 5 years ago

I trained with 49 pictures I took at high res using the command flow --model cfg/meu.cfg --load bin/yolo.weights --train --annotation train --dataset train --savepb --batch 1 --gpu 0.7 --load -1

I have just one class (a deodorant), which I want the system to identify in the camera

My labels.txt contains just line line: Old Spice

The XML files are like this: **<?xml version="1.0" ?>

images IMG_20181101_112452.jpg /root/darkflow/images/IMG_20181101_112452.jpg Unknown 3456 4608 3 0 Old Spice Unspecified 0 0 152 768 872 3992 2288 3952 2512 432 1512 216 904 328 ** My cfg file is : **[net] batch=1 subdivisions=1 width=416 height=416 channels=3 momentum=0.9 decay=0.0005 angle=0 saturation = 1.5 exposure = 1.5 hue=.1 learning_rate=0.001 max_batches = 10 policy=steps steps=-1,100,20000,30000 scales=.1,10,.1,.1 [convolutional] batch_normalize=1 filters=16 size=3 stride=1 pad=1 activation=leaky [maxpool] size=2 stride=2 [convolutional] batch_normalize=1 filters=32 size=3 stride=1 pad=1 activation=leaky [maxpool] size=2 stride=2 [convolutional] batch_normalize=1 filters=64 size=3 stride=1 pad=1 activation=leaky [maxpool] size=2 stride=2 [convolutional] batch_normalize=1 filters=128 size=3 stride=1 pad=1 activation=leaky [maxpool] size=2 stride=2 [convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky [maxpool] size=2 stride=2 [convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky [maxpool] size=2 stride=1 [convolutional] batch_normalize=1 filters=1024 size=3 stride=1 pad=1 activation=leaky ########### [convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=1024 activation=leaky [convolutional] size=1 stride=1 pad=1 filters=30 activation=linear [region] anchors = 1.08,1.19, 3.42,4.41, 6.63,11.38, 9.42,5.11, 16.62,10.52 bias_match=1 classes=1 coords=4 num=5 softmax=1 jitter=.2 rescore=1 object_scale=5 noobject_scale=1 class_scale=1 coord_scale=1 absolute=1 thresh = .5 random=1** After the train, where it would not below 2%, I exported the train using the command: **flow --model cfg/meu.cfg --load -1 --savepb --gpu 1.0** Finally, I ran the train using the command: **flow --pbLoad built_graph/meu.pb --metaLoad built_graph/meu.meta --demo camera** It does not recognize the deodorant: it just shows what the camera is getting and no boxes of any kind. Any ideas?
HSHunterR commented 5 years ago

You can turn down the threshold(--threshold 0.1 or 0.2) to see are there have bounding boxes.