fresh man ask for help about the dataset

[OliverHxh]

hi~~~ i am a new man in this. your repository is amazing for me, but i have a problem with the dataset.maybe its not a big problem for u. i already download the Pelee_VOC.pth and put it to weights folder, and i follow the steps to download the dataset VOC2007, and it looks like this. and i run the command python demo.py --dataset VOC\VOC2007 --config ./configs/Pelee_VOC.py --trained_model ./weights/Pelee_VOC.pth --show but it comes out the i dont whats wrong? could u help me? thx~~~

[OliverHxh]

em.... worong folder path i was so stupid...

[Z-Z-Han]

@Jaychoukobbe Hi! I also meet a similar KeyError. Could you tell me what the right way to set folder path is? Currently I put VOC data at home: ~/data/VOCdevkit and ~/coco.

[OliverHxh]

@Z-Z-Han if u wanto train on VOC, just use VOC is ok , because in Pelee_VOC.py file, this shows the dataset path, that all u need to do is to tell it what dataset u wanna train on VOC or COCO

[Z-Z-Han]

@Jaychoukobbe Get it! Thank you very very much!

[OliverHxh]

@Z-Z-Han i am still working on this codebase, if u have any question, u are welcome to ask...

[zhuixunforever]

@Jaychoukobbe @Z-Z-Han @yxlijun did you train your own dataset? or train coco/voc dataset by your self? did you success with this code? and how long had you been training??

[OliverHxh]

@zhuixunforever i replace backbone with shuffleNetV2 and i train on VOC, i use 1080i gpu and almost train 3days, 400 epoch. by the way my batchsize is 64. the final MAP result is 0.62 when i use shufflenetV2 1X, and 0.67 when i use shufflenetV2 2X

[Z-Z-Han]

@zhuixunforever I trained the model on my dataset with over one hundred images,using 1080 gpu. The time for training is about on day, and I only change batchsize to 32. The performance is a little worse than ssd with vgg as backbone, but also acceptable. For an image with size 640*480, the inferencing time is about 1~2 seconds when using cpu.

[OliverHxh]

@Z-Z-Han what dataset u train on? and waht map u get?

[Z-Z-Han]

My own dataset

---Original--- From: "Jaychoukobbe"[email protected] Date: Wed, Aug 7, 2019 17:17 PM To: "yxlijun/Pelee.Pytorch"[email protected]; Cc: "Z-Z-Han"[email protected];"Mention"[email protected]; Subject: Re: [yxlijun/Pelee.Pytorch] fresh man ask for help about the dataset (#4)

@Z-Z-Han what dataset u train on? and waht map u get?

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

[Z-Z-Han]

Not public available

---Original--- From: "Jaychoukobbe"[email protected] Date: Wed, Aug 7, 2019 17:17 PM To: "yxlijun/Pelee.Pytorch"[email protected]; Cc: "Z-Z-Han"[email protected];"Mention"[email protected]; Subject: Re: [yxlijun/Pelee.Pytorch] fresh man ask for help about the dataset (#4)

@Z-Z-Han what dataset u train on? and waht map u get?

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

[OliverHxh]

@Z-Z-Han have ever tried on public dataset , like voc or coco?

[Z-Z-Han]

Not yet

---Original--- From: "Jaychoukobbe"[email protected] Date: Wed, Aug 7, 2019 17:21 PM To: "yxlijun/Pelee.Pytorch"[email protected]; Cc: "Z-Z-Han"[email protected];"Mention"[email protected]; Subject: Re: [yxlijun/Pelee.Pytorch] fresh man ask for help about the dataset (#4)

@Z-Z-Han have ever tried on public dataset , like voc or coco?

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

[OliverHxh]

@Z-Z-Han all right

[duducode]

@zhuixunforever I trained the model on my dataset with over one hundred images,using 1080 gpu. The time for training is about on day, and I only change batchsize to 32. The performance is a little worse than ssd with vgg as backbone, but also acceptable. For an image with size 640*480, the inferencing time is about 1~2 seconds when using cpu.

hao can i running it on cup,can you help me

[Rafay2919]

@Z-Z-Han i am still working on this codebase, if u have any question, u are welcome to ask...

i have following error "ImportError: No module named cpu_nms" how did you solve this.