video-parameter.txt

https://github.com/Cartucho/OpenLabeling

http://rlc-511w.joe.lan/cgi-bin/
https://home.jru.me/bee-cam/api.cgi?cmd=Snap&channel=0&rs=sdilj23SDO3DDGHJsdfs&user=guest&password=my_guest&1555017246

darknet detector test bee-counter.data bwh.cfg backup/bwh_last.weights -thresh 0.25 api.jpeg
darknet detector test bee-counter.data bwh2.cfg backup/bwh2_last.weights -thresh 0.15 api.jpeg


darknet detector demo bee-counter.data bwh2.cfg backup/bwh2_last.weights -thresh 0.15 rtsp://guest:my_guest@rlc-511w.joe.lan:554/

convert rtsp to http-live-stream
https://medium.com/androvideo/convert-rtsp-to-hls-using-ffmpeg-2fe2cdf3a0de

ffmpeg -fflags nobuffer \
 -rtsp_transport tcp \
 -i rtsp://guest:my_guest@rlc-511w.joe.lan:554/ \
 -vsync 0 \
 -copyts \
 -vcodec copy \
 -movflags frag_keyframe+empty_moov \
 -an \
 -hls_flags delete_segments+append_list \
 -f segment \
 -segment_list_flags live \
 -segment_time 1 \
 -segment_list_size 3 \
 -segment_format mpegts \
 -segment_list /tmp/stream/index.m3u8 \
 -segment_list_type m3u8 \
 -segment_list_entry_prefix /stream/ \
 /tmp/stream/%d.ts



Get-ChildItem ".\data\batch5\" -Filter *.jpg |
Foreach-Object {
	& darknet.exe detector test bee-counter.data bwh2.cfg backup/bwh2_last.weights -thresh 0.25 -dont_show -save_labels
}

darknet partial bwh.cfg backup/bwh_last.weights backup/bwh_last.weights 74
darknet partial bwh2.cfg backup/yv3t_last.weights backup/bwh2_last.weights 74

darknet partial yolov3_spp.cfg yolov3-spp.weights backup/yolov3_spp_last.weights 74




To calculate anchors: 
darknet.exe detector calc_anchors bee-counter.data -num_of_clusters 9 -width 608 -height 608
------------
bwh2

 num_of_clusters = 9, width = 1024, height = 768
 read labels from 1049 images
 loaded          image: 1049     box: 6388
 all loaded.

 calculating k-means++ ...

 iterations = 137


 avg IoU = 76.35 %

Saving anchors to the file: anchors.txt
anchors =  40, 40,  62, 48,  54, 76,  86, 73, 146,154, 261,259, 476,323, 465,503, 721,533

--------------
bwh
darknet.exe detector calc_anchors bee-counter.data -num_of_clusters 9 -width 768 -height 576

 num_of_clusters = 9, width = 768, height = 576
 read labels from 1007 images
 loaded          image: 1007     box: 5893
 all loaded.

 calculating k-means++ ...

 iterations = 91


 avg IoU = 76.32 %

Saving anchors to the file: anchors.txt
anchors =  29, 30,  45, 36,  41, 58,  64, 54, 111,118, 200,196, 366,236, 347,371, 540,402