ameyawagh / traffic_sign_detection Goto Github PK

View Code? Open in Web Editor NEW

71.0 8.0 21.0 40.22 MB

Implementation of darkflow on traffic sign detection and classification

License: MIT License

Python 99.68% Shell 0.32%

yolo traffic-sign-classification traffic-sign-detection tensorflow

traffic_sign_detection's Introduction

Traffic_sign_detection_YOLO

Detecting traffic signs using YOLO algorithm

Clone the repository

git clone https://github.com/AmeyaWagh/Traffic_sign_detection_YOLO.git

Goto darkflow and build cython extension by running

cd darkflow
python3 setup.py build_ext --inplace

Then build globally with

pip install .

Check if "flow" works with "flow --h"

flow --h

Go back and create a new folder called "dataset" in base directory. Download and extract LISA dataset into the dataset folder

cd ..
mkdir dataset

run datasetGenerator.py

python3 datasetGenerator.py

goto darkflow and create "built_graph" directory inside darkflow if you are not training, and save pb and meta files there (pb and meta files can be downloaded here "https://drive.google.com/file/d/171AyNg4zSmz4OXhfcdgU2cxrqTfIV2BD/view?usp=sharing")

cd darkflow
mkdir built_graph

set GPU to 0.0 in the config3.json if not using GPU

{
	"yoloConfig":{
		"pbLoad": "./built_graph/tiny-yolo-voc27.pb", 
		"metaLoad": "./built_graph/tiny-yolo-voc27.meta",
		"labels":"../labels.txt",
		"threshold":0.01, 
		"gpu":0.7
	},
	"dataset":"./dataset"	
}

Run YOLO

./runYOLO

Training

cd darkflow
./trainYOLO

traffic_sign_detection's People

Contributors

Stargazers

Watchers

traffic_sign_detection's Issues

Unable to test pretrained weights with my own data

Hi amey,
Thanks for sharing work, I tried detection with your data but with my data prediction gives empty array. Even I keep image size same as yours. Please let me know what changes should I make in my image data.
Thanks,
Tanaji

Permission denied: './ckpt/checkpoint'

Traceback (most recent call last):
File "./flow", line 6, in
cliHandler(sys.argv)
File "C:\SDCPD\Traffic_sign_detection_YOLO-master\Traffic_sign_detection_YOLO-master\darkflow\darkflow\cli.py", line 26, in cliHandler
tfnet = TFNet(FLAGS)
File "C:\SDCPD\Traffic_sign_detection_YOLO-master\Traffic_sign_detection_YOLO-master\darkflow\darkflow\net\build.py", line 76, in init
self.setup_meta_ops()
File "C:\SDCPD\Traffic_sign_detection_YOLO-master\Traffic_sign_detection_YOLO-master\darkflow\darkflow\net\build.py", line 151, in setup_meta_ops
if self.FLAGS.load != 0: self.load_from_ckpt()
File "C:\SDCPD\Traffic_sign_detection_YOLO-master\Traffic_sign_detection_YOLO-master\darkflow\darkflow\net\help.py", line 23, in load_from_ckpt
with open(os.path.join(self.FLAGS.backup, 'checkpoint'), 'r') as f:
PermissionError: [Errno 13] Permission denied: './ckpt/checkpoint'

even tried running cmd promt as administrator

Stuck on ./runYOLO

I ran it on Raspberry PI 3b. Previous steps are fine, but I stuck on "./runYOLO". I tried "sudo ./runYOLO", but still the same:

{'dataset': '/home/pi/Traffic_sign_detection_YOLO/dataset', 'yoloConfig': {'metaLoad': '/home/pi/Traffic_sign_detection_YOLO/darkflow/built_graph/tiny-yolo-voc27.meta', 'threshold': 0.01, 'pbLoad': '/home/pi/Traffic_sign_detection_YOLO/darkflow/built_graph/tiny-yolo-voc27.pb', 'gpu': 0.0, 'labels': '../labels.txt'}}

Loading from .pb and .meta
Traceback (most recent call last):
File "YOLOtest.py", line 126, in
det = objectDetector(video=False)
File "YOLOtest.py", line 21, in init
self.tfnet = TFNet(self.options)
File "/home/pi/Traffic_sign_detection_YOLO/darkflow/darkflow/net/build.py", line 54, in init
self.build_from_pb()
File "/home/pi/Traffic_sign_detection_YOLO/darkflow/darkflow/net/build.py", line 87, in build_from_pb
name=""
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/importer.py", line 258, in import_graph_def
op_def = op_dict[node.op]
KeyError: 'RealDiv'

Please help me, thx

Inquiry about Dataset Source and Access

Hello,

I am reaching out to inquire about the source of the dataset used in the Traffic_sign_detection_YOLO repository. I am interested in accessing the dataset for further analysis and research purposes. Could you please provide information regarding the dataset source and the process for obtaining access to the dataset?

Thank you for your assistance.

Best regards,
yihong1120

permission denied even with sudo command

I tried by different means to copy the meta and pb files, but with no success. I get permission denied even with sudo command. Can you please help me.

how to run the code for a custom video/images

hi Ameya ,
the project you have done looks cool. I was trying to test it on my system. Everything is working fine, I executed the ./runyolo file and got the pop up of the detected objects for all the training images. I would like to know where to change the code(to add the path of my test video/test images) such that I could execute the detection on my test video as you have done for a worcester video.

Is this the code that we need to make changes for in YOLOtest.py, I tried to change the line where
self.cap = cv2.VideoCapture('../../WPI_vdo.mov') and put my video, but I was unable to process, what could be the problem.

Thanks in advance :)