output picture have so many bounding box, no NMS.

[GUOXiangX]

I run the command lines :

1. "python flow --imgdir sample_img/ --model cfg/yolo-tiny.cfg --load bin/yolo-tiny.weights --gpu 1.00"
2. "python flow --imgdir sample_img/ --model cfg/yolo-tiny.cfg --load bin/yolo-tiny.weights --json"

,but the json output and predictions include all bounding box, having no NMS.So I hope finding a solution.

[GUOXiangX]

Thanks，this problem has been solved with your help.

                                邮箱：14738427@qq.com

签名由 网易邮箱大师 定制 On 02/22/2018 23:22, David Forester wrote: could you try detections on an image using a few different thresholds such as 0.3, 0.5., and 0.7, to see what the differences are? Like below: $ python flow --imgdir sample_img/ --model cfg/yolo-tiny.cfg --load bin/yolo-tiny.weights --gpu 1.00 --threshold 0.5

—You are receiving this because you authored the thread.Reply to this email directly, view it on GitHub, or mute the thread.

{"api_version":"1.0","publisher":{"api_key":"05dde50f1d1a384dd78767c55493e4bb","name":"GitHub"},"entity":{"external_key":"github/thtrieu/darkflow","title":"thtrieu/darkflow","subtitle":"GitHub repository","main_image_url":"https://cloud.githubusercontent.com/assets/143418/17495839/a5054eac-5d88-11e6-95fc-7290892c7bb5.png","avatar_image_url":"https://cloud.githubusercontent.com/assets/143418/15842166/7c72db34-2c0b-11e6-9aed-b52498112777.png","action":{"name":"Open in GitHub","url":"https://github.com/thtrieu/darkflow"}},"updates":{"snippets":[{"icon":"PERSON","message":"@drforester in #590: could you try detections on an image using a few different thresholds such as 0.3, 0.5., and 0.7, to see what the differences are? Like below: \r\n$ python flow --imgdir sample_img/ --model cfg/yolo-tiny.cfg --load bin/yolo-tiny.weights --gpu 1.00 --threshold 0.5 "}],"action":{"name":"View Issue","url":"https://github.com/thtrieu/darkflow/issues/590#issuecomment-367714973"}}}

[HebaAladdin]

Can you tell me what did you do to solve this ?

[GUOXiangX]

could you try detections on an image using a few different thresholds such as 0.3, 0.5., and 0.7, to see what the differences are? Like below: $ python flow --imgdir sample_img/ --model cfg/yolo-tiny.cfg --load bin/yolo-tiny.weights --gpu 1.00 --threshold 0.5

——answer by David Forester

                                邮箱：14738427@qq.com

签名由 网易邮箱大师 定制 On 03/25/2018 19:47, HebaAladdin wrote: Can you tell me what did you do to solve this ?

—You are receiving this because you authored the thread.Reply to this email directly, view it on GitHub, or mute the thread.

{"api_version":"1.0","publisher":{"api_key":"05dde50f1d1a384dd78767c55493e4bb","name":"GitHub"},"entity":{"external_key":"github/thtrieu/darkflow","title":"thtrieu/darkflow","subtitle":"GitHub repository","main_image_url":"https://cloud.githubusercontent.com/assets/143418/17495839/a5054eac-5d88-11e6-95fc-7290892c7bb5.png","avatar_image_url":"https://cloud.githubusercontent.com/assets/143418/15842166/7c72db34-2c0b-11e6-9aed-b52498112777.png","action":{"name":"Open in GitHub","url":"https://github.com/thtrieu/darkflow"}},"updates":{"snippets":[{"icon":"PERSON","message":"@HebaAladdin in #590: Can you tell me what did you do to solve this ?\r\n"}],"action":{"name":"View Issue","url":"https://github.com/thtrieu/darkflow/issues/590#issuecomment-375964314"}}}

[GUOXiangX]

        could you try detections on an image using a few different thresholds such as 0.3, 0.5., and 0.7, to see what the differences are? Like below:$ python flow --imgdir sample_img/ --model cfg/yolo-tiny.cfg --load bin/yolo-tiny.weights --gpu 1.00 --threshold 0.5——answer by David Forester邮箱：14738427@qq.com签名由 网易邮箱大师 定制

                                邮箱：14738427@qq.com

签名由 网易邮箱大师 定制 On 03/25/2018 19:47, HebaAladdin wrote: Can you tell me what did you do to solve this ?

—You are receiving this because you authored the thread.Reply to this email directly, view it on GitHub, or mute the thread.

{"api_version":"1.0","publisher":{"api_key":"05dde50f1d1a384dd78767c55493e4bb","name":"GitHub"},"entity":{"external_key":"github/thtrieu/darkflow","title":"thtrieu/darkflow","subtitle":"GitHub repository","main_image_url":"https://cloud.githubusercontent.com/assets/143418/17495839/a5054eac-5d88-11e6-95fc-7290892c7bb5.png","avatar_image_url":"https://cloud.githubusercontent.com/assets/143418/15842166/7c72db34-2c0b-11e6-9aed-b52498112777.png","action":{"name":"Open in GitHub","url":"https://github.com/thtrieu/darkflow"}},"updates":{"snippets":[{"icon":"PERSON","message":"@HebaAladdin in #590: Can you tell me what did you do to solve this ?\r\n"}],"action":{"name":"View Issue","url":"https://github.com/thtrieu/darkflow/issues/590#issuecomment-375964314"}}}

[GUOXiangX]

could you try detections on an image using a few different thresholds such as 0.3, 0.5., and 0.7, to see what the differences are? Like below: $ python flow --imgdir sample_img/ --model cfg/yolo-tiny.cfg --load bin/yolo-tiny.weights --gpu 1.00 --threshold 0.5

——answer by David Forester

                        14738427

                                邮箱：14738427@qq.com

签名由 网易邮箱大师 定制 On 03/25/2018 19:47, HebaAladdin wrote: Can you tell me what did you do to solve this ?

—You are receiving this because you authored the thread.Reply to this email directly, view it on GitHub, or mute the thread.

{"api_version":"1.0","publisher":{"api_key":"05dde50f1d1a384dd78767c55493e4bb","name":"GitHub"},"entity":{"external_key":"github/thtrieu/darkflow","title":"thtrieu/darkflow","subtitle":"GitHub repository","main_image_url":"https://cloud.githubusercontent.com/assets/143418/17495839/a5054eac-5d88-11e6-95fc-7290892c7bb5.png","avatar_image_url":"https://cloud.githubusercontent.com/assets/143418/15842166/7c72db34-2c0b-11e6-9aed-b52498112777.png","action":{"name":"Open in GitHub","url":"https://github.com/thtrieu/darkflow"}},"updates":{"snippets":[{"icon":"PERSON","message":"@HebaAladdin in #590: Can you tell me what did you do to solve this ?\r\n"}],"action":{"name":"View Issue","url":"https://github.com/thtrieu/darkflow/issues/590#issuecomment-375964314"}}}

[HebaAladdin]

@GUOXiang the answer isn't correctly displayed I can't see it :)