Complex-YOLOv4-Pytorch:本文基于YOLOv4的PyTorch实现

# Complex YOLOv4 [![python-image]][python-url] [![pytorch-image]][pytorch-url] The PyTorch Implementation based on YOLOv4 of the paper: [Complex-YOLO: Real-time 3D Object Detection on Point Clouds](https://arxiv.org/pdf/1803.06199.pdf) --- ## Features - [x] Realtime 3D object detection based on YOLOv4 - [x] Support [distributed data parallel training](https://github.com/pytorch/examples/tree/master/distributed/ddp) - [x] Tensorboard - [x] Mosaic/Cutout augmentation for training - [x] Use [GIoU](https://arxiv.org/pdf/1902.09630v2.pdf) loss of rotated boxes for optimization. - **Update 2020.08.26**: [Super Fast and Accurate 3D Object Detection based on 3D LiDAR Point Clouds](https://github.com/maudzung/Super-Fast-Accurate-3D-Object-Detection) - Faster training, faster inference - An Anchor-free approach - No need for Non-Max-Suppression - Demonstration (on a GTX 1080Ti) [](http://www.youtube.com/watch?v=FI8mJIXkgX4) **[Youtube link](https://youtu.be/FI8mJIXkgX4)** ## 2. Getting Started ### 2.1. Requirement ```shell script pip install -U -r requirements.txt ``` For [`mayavi`](https://docs.enthought.com/mayavi/mayavi/installation.html) and [`shapely`](https://shapely.readthedocs.io/en/latest/project.html#installing-shapely) libraries, please refer to the installation instructions from their official websites. ### 2.2. Data Preparation Download the 3D KITTI detection dataset from [here](http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d). The downloaded data includes: - Velodyne point clouds _**(29 GB)**_: input data to the Complex-YOLO model - Training labels of object data set _**(5 MB)**_: input label to the Complex-YOLO model - Camera calibration matrices of object data set _**(16 MB)**_: for visualization of predictions - Left color images of object data set _**(12 GB)**_: for visualization of predictions Please make sure that you construct the source code & dataset directories structure as below. For 3D point cloud preprocessing, please refer to the previous works: - [VoxelNet-Pytorch](https://github.com/skyhehe123/VoxelNet-pytorch) - [Complex-YOLOv2](https://github.com/AI-liu/Complex-YOLO) - [Complex-YOLOv3](https://github.com/ghimiredhikura/Complex-YOLOv3) ### 2.3. Complex-YOLO architecture  This work has been based on the paper [YOLOv4: Optimal Speed and Accuracy of Object Detection](https://arxiv.org/abs/2004.10934). Please refer to several implementations of YOLOv4 using PyTorch DL framework: - [Tianxiaomo/pytorch-YOLOv4](https://github.com/Tianxiaomo/pytorch-YOLOv4) - [Ultralytics/yolov3_and_v4](https://github.com/ultralytics/yolov3) - [WongKinYiu/PyTorch_YOLOv4](https://github.com/WongKinYiu/PyTorch_YOLOv4) - [VCasecnikovs/Yet-Another-YOLOv4-Pytorch](https://github.com/VCasecnikovs/Yet-Another-YOLOv4-Pytorch) ### 2.4. How to run #### 2.4.1. Visualize the dataset (both BEV images from LiDAR and camera images) ```shell script cd src/data_process ``` - To visualize BEV maps and camera images (with 3D boxes), let's execute _**(the `output-width` param can be changed to show the images in a bigger/smaller window)**_: ```shell script python kitti_dataloader.py --output-width 608 ``` - To visualize mosaics that are composed from 4 BEV maps (Using during training only), let's execute: ```shell script python kitti_dataloader.py --show-train-data --mosaic --output-width 608 ``` By default, there is _**no padding**_ for the output mosaics, the feature could be activated by executing: ```shell script python kitti_dataloader.py --show-train-data --mosaic --random-padding --output-width 608 ``` - To visualize cutout augmentation, let's execute: ```shell script python kitti_dataloader.py --show-train-data --cutout_prob 1. --cutout_nholes 1 --cutout_fill_value 1. --cutout_ratio 0.3 --output-width 608 ``` #### 2.4.2. Inference Download the trained model from [**_here_**](https://drive.google.com/drive/folders/1RHD9PBvk-9SjbKwoi_Q1kl9-UGFo2Pth?usp=sharing), then put it to `${ROOT}/checkpoints/` and execute: ```shell script python test.py --gpu_idx 0 --pretrained_path ../checkpoints/complex_yolov4/complex_yolov4_mse_loss.pth --cfgfile ./config/cfg/complex_yolov4.cfg --show_image ``` #### 2.4.3. Evaluation ```shell script python evaluate.py --gpu_idx 0 --pretrained_path <PATH> --cfgfile <CFG> --img_size <SIZE> --conf-thresh <THRESH> --nms-thresh <THRESH> --iou-thresh <THRESH> ``` (The `conf-thresh`, `nms-thresh`, and `iou-thresh` params can be adjusted. By default, these params have been set to _**0.5**_) #### 2.4.4. Training ##### 2.4.4.1. Single machine, single gpu ```shell script python train.py --gpu_idx 0 --batch_size <N> --num_workers <N>... ``` ##### 2.4.4.2. Multi-processing Distributed Data Parallel Training We should always use the `nccl` backend for multi-processing distributed training since it currently provides the best distributed training performance. - **Single machine (node), multiple GPUs** ```shell script python train.py --dist-url 'tcp://127.0.0.1:29500' --dist-backend 'nccl' --multiprocessing-distributed --world-size 1 --rank 0 ``` - **Two machines (two nodes), multiple GPUs** _**First machine**_ ```shell script python train.py --dist-url 'tcp://IP_OF_NODE1:FREEPORT' --dist-backend 'nccl' --multiprocessing-distributed --world-size 2 --rank 0 ``` _**Second machine**_ ```shell script python train.py --dist-url 'tcp://IP_OF_NODE2:FREEPORT' --dist-backend 'nccl' --multiprocessing-distributed --world-size 2 --rank 1 ``` To reproduce the results, you can run the bash shell script ```bash ./train.sh ``` #### Tensorboard - To track the training progress, go to the `logs/` folder and ```shell script cd logs/<saved_fn>/tensorboard/ tensorboard --logdir=./ ``` - Then go to [http://localhost:6006/](http://localhost:6006/): ### 2.5. List of usage for Bag of Freebies (BoF) & Bag of Specials (BoS) in this implementation | |Backbone | Detector | |---|---|---| |**BoF** |[x] Dropblock <br> [x] Random rescale, rotation (global) <br> [x] Mosaic/Cutout augmentation|[x] Cross mini-Batch Normalization <br>[x] Dropblock <br> [x] Random training shapes <br> | |**BoS** |[x] Mish activation <br> [x] Cross-stage partial connections (CSP) <br> [x] Multi-input weighted residual connections (MiWRC) |[x] Mish activation <br> [x] SPP-block <br> [x] SAM-block <br> [x] PAN path-aggregation block <br> [x] GIoU loss <br> [ ] CIoU loss | ## Contact If you think this work is useful, please give me a star! <br> If you find any errors or have any suggestions, please contact me (**Email:** `nguyenmaudung93.kstn@gmail.com`). <br> Thank you! ## Citation ```bash @article{Complex-YOLO, author = {Martin Simon, Stefan Milz, Karl Amende, Horst-Michael Gross}, title = {Complex-YOLO: Real-time 3D Object Detection on Point Clouds}, year = {2018}, journal = {arXiv}, } @article{YOLOv4, author = {Alexey Bochkovskiy, Chien-Yao Wang, Hong-Yuan Mark Liao}, title = {YOLOv4: Optimal Speed and Accuracy of Object Detection}, year = {2020}, journal = {arXiv}, } ``` ## Folder structure ``` ${ROOT} └── checkpoints/ ├── complex_yolov3/ └── complex_yolov4/ └── dataset/ └── kitti/ ├──ImageSets/ │ ├── train.txt │ └── val.txt ├── training/ │ ├── image_2/ <-- for visualization │ ├── calib/ │ ├── label_2/ │ └── velodyne/ └── testing/ │ ├── image_2/ <-- for visualization │ ├── calib/ │ └── velodyne/ └── classes_names.txt └── src/ ├── config/ ├── cfg/ │   ├── complex_yolov3.cfg │   ├── complex_yolov3_tiny.cfg │   ├�