deep_sort_pytorch

deep_sort_pytorch copied to clipboard

Hi, I find that NMS has been used twice. First, to detect the object bbox, yolov3 employs the NMS on the all detected objects, and return `bbox_xywh`, `cls_conf` and `cls_ids`,...

siyilingting

Tracking using provided detections

2

Hello, happy new year! Thanks for sharing the codes. Can I use these codes to track my own detections? Because I have obtained the detections from the front-end detector, so...

EddieEduardo

Model Differences From The Reference Paper

2

I know that you made some modifications on the vanilla deep sort and your repo is working just fine. But as i checked the paper for deep sort, saw that...

ghost

After installing `mmdet` and `mmcv` by myself, an error occurs: > ModuleNotFoundError: No module named 'mmcv._ext This error for `mmdet==0.6.0` and `mmcv==0.6.2`: > ImportError: cannot import name 'deform_conv_cuda' Enviroment: -...

ZhangYuef

size mismatch for classifier.4.weight: copying a param with shape torch.Size([751, 256]) from checkpoint, the shape in current model is torch.Size([1, 256]).

11

I still do not understand why we need 751 classes for just extractor feature of persons? ``` ~/Desktop/deep_sort_pytorch/deep_sort/deep$ python train.py --resume Loading from checkpoint/ckpt.t7 Traceback (most recent call last): File...

alexnguyen2201

Error when changing the torchvision.transforms.RandomCrop((128,64),padding=4) to Resize when training Deep network

1

I have changed the `torchvision.transforms.RandomCrop((128,64),padding=4)` to `torchvision.transforms.Resize((128,64))` as in [transform_test](https://github.com/ZQPei/deep_sort_pytorch/blob/4c2d86229b0b69316af67d519f8476eee69c9b20/deep_sort/deep/train.py#L37) (because my image size is small) And when I changed it to square, there's problem appears: `RuntimeError: mat1 and mat2...

ithmz

for some video frames throws above exception from `bbox_tlwh[:,0] = bbox_xywh[:,0] - bbox_xywh[:,2]/2.` line

dulangaheshan

Evaluation Results with MOT16

12

Dear @ZQPei I evaluated the performance using MOT16 training datasets and got the following result, just wondering is it reasonable? IDF1 IDP IDR Rcll Prcn GT MT PT ML FP...

Mary-xl

I cann't build nms on windows.

1

I try to build the nms on windows. It send me an error that shows below.    This is my library details. 

beebrain

When a new object appears in the field of view，the box predicted is much larger.Is the problem in yolo or feature extractor

1

    the white box is predicted by yolo,color box is predicted by Kalman filter

tdf1995