Panoptic driving Perception (YOLOPv2)

https://github.com/CAIC-AD/YOLOPv2

https://arxiv.org/abs/2208.11434
YOLOPv2: Better, Faster, Stronger for Panoptic Driving Perception

 

0. Rosbag to mp4

sip2@sip2-2021:~/catkin_od/src/object_detection/scripts/YOLOPv2$ cd ~
sip2@sip2-2021:~$ cd sample_code/
PythonRobotics/ rosbag2video/   
sip2@sip2-2021:~/sample_code/rosbag2video$ ./rosbag2video.py --topic /camera/color/image_raw /media/sip2/SIP2022/2022-09-16-iam-5.bag
############# UNCOMPRESSED IMAGE ######################
/camera/color/image_raw  with datatype: sensor_msgs/Image

finished653 fps=297 q=28.0 size=   27392kB time=00:05:43.64 bitrate= 653.0kbits/s speed=11.8x    
frame= 8798 fps=294 q=-1.0 Lsize=   28093kB time=00:05:51.80 bitrate= 654.2kbits/s speed=11.7x  
 

 

1. YOLOPv2

sip2@sip2-2021:~$ source ~/anaconda3/etc/profile.d/conda.sh
sip2@sip2-2021:~$ conda info --e
# conda environments:
#
base                  *  /home/sip2/anaconda3
py38-test                /home/sip2/anaconda3/envs/py38-test
py38-torch1-12-1         /home/sip2/anaconda3/envs/py38-torch1-12-1
py38-torch1-12-1-gpu-od     /home/sip2/anaconda3/envs/py38-torch1-12-1-gpu-od

sip2@sip2-2021:~$ conda create -n py38-torch1-12-1-gpu-yolopv2 --clone py38-torch1-12-1
Source:      /home/sip2/anaconda3/envs/py38-torch1-12-1
Destination: /home/sip2/anaconda3/envs/py38-torch1-12-1-gpu-yolopv2
Packages: 65
Files: 0
Preparing transaction: done
Verifying transaction: |
SafetyError: The package for pytorch located at /home/sip2/anaconda3/pkgs/pytorch-1.12.1-py3.8_cuda11.6_cudnn8.3.2_0
appears to be corrupted. The path 'lib/python3.8/site-packages/torch/nn/modules/upsampling.py'
has an incorrect size.
  reported size: 11056 bytes
  actual size: 11005 bytes


done
Executing transaction: - By downloading and using the CUDA Toolkit conda packages, you accept the terms and conditions of the CUDA End User License Agreement (EULA): https://docs.nvidia.com/cuda/eula/index.html

done
#
# To activate this environment, use
#
#     $ conda activate py38-torch1-12-1-gpu-yolopv2
#
# To deactivate an active environment, use
#
#     $ conda deactivate

(py38-torch1-12-1-gpu-yolopv2) sip2@sip2-2021:~/catkin_od/src/object_detection/scripts$ git clone https://github.com/CAIC-AD/YOLOPv2.git
Cloning into 'YOLOPv2'...
remote: Enumerating objects: 162, done.
remote: Counting objects: 100% (49/49), done.
remote: Compressing objects: 100% (44/44), done.
remote: Total 162 (delta 34), reused 6 (delta 5), pack-reused 113
Receiving objects: 100% (162/162), 57.29 MiB | 8.22 MiB/s, done.
Resolving deltas: 100% (60/60), done.

(py38-torch1-12-1-gpu-yolopv2) sip2@sip2-2021:~/catkin_od/src/object_detection/scripts$ cd YOLOPv2/

 

(py38-torch1-12-1-gpu-yolopv2) sip2@sip2-2021:~/catkin_od/src/object_detection/scripts/YOLOPv2$ pip install -r requirements.txt

# Usage: pip install -r requirements.txt

# Base ----------------------------------------

matplotlib>=3.2.2

numpy>=1.18.5

opencv-python>=4.1.1

Pillow>=7.1.2

PyYAML>=5.3.1

requests>=2.23.0

scipy>=1.4.1

torch>=1.7.0

torchvision>=0.8.1,!=0.13.0

tqdm>=4.41.0

protobuf<4.21.3

# Logging -------------------------------------

tensorboard>=2.4.1

# wandb

# Plotting ------------------------------------

pandas>=1.1.4

seaborn>=0.11.0

# Export --------------------------------------

# coremltools>=4.1 # CoreML export

# onnx>=1.9.0 # ONNX export

# onnx-simplifier>=0.3.6 # ONNX simplifier

# scikit-learn==0.19.2 # CoreML quantization

# tensorflow>=2.4.1 # TFLite export

# tensorflowjs>=3.9.0 # TF.js export

# openvino-dev # OpenVINO export

# Extras --------------------------------------

ipython # interactive notebook

psutil # system utilization

thop # FLOPs computation

# albumentations>=1.0.3

# pycocotools>=2.0 # COCO mAP

# roboflow

(py38-torch1-12-1-gpu-yolopv2) sip2@sip2-2021:~/catkin_od/src/object_detection/scripts/YOLOPv2$ python demo.py  --source data/example.jpg

Namespace(agnostic_nms=False, classes=None, conf_thres=0.3, device='0', exist_ok=False, img_size=640, iou_thres=0.45, name='exp', nosave=False, project='runs/detect', save_conf=False, save_txt=False, source='data/example.jpg', weights='data/weights/yolopv2.pt')
/home/sip2/anaconda3/envs/py38-torch1-12-1-gpu-yolopv2/lib/python3.8/site-packages/torch/functional.py:478: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at  /opt/conda/conda-bld/pytorch_1659484683044/work/aten/src/ATen/native/TensorShape.cpp:2894.)
  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
384x640 Done. (1.498s)
 The image with the result is saved in: runs/detect/exp/example.jpg
inf : (1.4985s/frame)   nms : (0.0059s/frame)
Done. (1.566s)