/CenterNet-plus

A Simple Baseline for Object Detection

Primary LanguagePython

CenterNet-plus

A Simple Baseline for Object Detection based on CenterNet with ResNet backbone.

CenterNet is a very simple yet efficient object detector. Based on this supreme work, I rebuild it with PyTorch.

CenterNet is an encoder-decoder network, but I won't consider Hourglass-101 in this project as it is heavy and time consuming.

I will try DLA-34 in the future, but let us focus ResNet for now.

Weight

You can download all my models from my BaiDuYunDisk:

Link: https://pan.baidu.com/s/1yaAhIT6NErzv_QaDHqfYrg

Password: zr8m

I will upload them to Google Drive.

Backbone

For backbone, I use ResNet including ResNet-18, ResNet-50 and ResNet-101. I also make use of DarkNet-53 and CSPDarkNet-53 reproduced by myself with PyTorch.

Neck

For neck, I use the DilateEncoder proposed by YOLOF. DCN or DCNv2 don't be considered as they are not easy to deploy.

Decoder

For decoder, I just use nearest interpolate following a convolutional layer rather than deconvolutional layer to avoid the effect of checkerboard.

Detection head

For detection head, there are totally 4 branches including class branch, offset(tx and ty) branch, size(tw and th) branch and iou-aware branch. During training stage, how to get the labels of offset and size is different from CenterNet. For more details, see my codes in tools.py file.

The whole structure of my CenterNet-plus is shown in the following picture:

Image

I dont deploy any DCN in my CenterNet-plus.

Train on COCO

For example, you can use the following line of command to train my CenterNet-plus with ResNet-18 backbone.

python train.py -d coco --cuda --ema -bk r18

Experimental results

COCO

Experimental results on COCO:

backbone data size AP AP50 AP75 AP_S AP_M AP_L
CenterNet-plus ResNet-18 COCO val 512x512 29.9 49.1 31.8 14.4 31.0 43.1
CenterNet-plus ResNet-50 COCO val 512x512 36.0 56.2 38.9 18.7 38.9 51.1
CenterNet-plus ResNet-101 COCO val 512x512 37.5 57.7 41.0 19.5 41.4 53.0
CenterNet ResNet-18 COCO val 512x512 28.1 44.9 29.6 - - -
CenterNet ResNet-101 COCO val 512x512 34.6 53.0 36.9 - - -
CenterNet DLA-34 COCO val 512x512 37.4 55.1 40.8 - - -
CenterNet Hourglass-104 COCO val 512x512 40.3 59.1 44.0 - - -

With ResNet backbone, my CenterNet-plus works better.

Things I tried that did not work

  • multi scale training
  • mosaic augmentation

When I use multi-scale training trick, my CenterNet-plus with ResNet18 got 25 AP lower than 29.9 AP.

When I use mosaic augmentation, my CenterNet-plus with ResNet18 got only 25 AP.

What I am going to do next?

  • Replace txtytwth by GIoU to learning regression.
  • Add more positive samples to accelerated training.
  • Large image size: 640.
  • Try DLA-34 backbone.
  • DDP training mode.