A Keras implementation of YOLOv3 (Tensorflow backend) for vehicle detection inspired by qqwweee.
- Download YOLOv3 weights from YOLO website.
- Convert the Darknet YOLO model to a Keras model.
- Run YOLO detection.
wget https://pjreddie.com/media/files/yolov3.weights
python convert.py yolov3.cfg yolov3.weights model_data/yolo.h5
python new_yolo_video_tiny.py [OPTIONS...] --image, for image detection mode, OR
python new_yolo_video_tiny.py [video_path] [output_path (optional)]
For Tiny YOLOv3, just do in a similar way, just specify model path and anchor path with --model model_file
and --anchors anchor_file
.
Use --help to see usage of yolo_video.py:
usage: yolo_video.py [-h] [--model MODEL] [--anchors ANCHORS]
[--classes CLASSES] [--gpu_num GPU_NUM] [--image]
[--input] [--output]
positional arguments:
--input Video input path
--output Video output path
optional arguments:
-h, --help show this help message and exit
--model MODEL path to model weight file, default model_data/yolo.h5
--anchors ANCHORS path to anchor definitions, default
model_data/yolo_anchors.txt
--classes CLASSES path to class definitions, default
model_data/coco_classes.txt
--gpu_num GPU_NUM Number of GPU to use, default 1
--image Image detection mode, will ignore all positional arguments
- MultiGPU usage: use
--gpu_num N
to use N GPUs. It is passed to the Keras multi_gpu_model().
-
Generate your own annotation file and class names file.
One row for one image;
Row format:image_file_path box1 box2 ... boxN
;
Box format:x_min,y_min,x_max,y_max,class_id
(no space).
For VOC dataset, trypython voc_annotation.py
Here is an example:path/to/img1.jpg 50,100,150,200,0 30,50,200,120,3 path/to/img2.jpg 120,300,250,600,2 ...
-
Make sure you have run
python convert.py -w yolov3.cfg yolov3.weights model_data/yolo_weights.h5
The file model_data/yolo_weights.h5 is used to load pretrained weights. -
Modify train.py and start training.
python train.py
Use your trained weights or checkpoint weights with command line option--model model_file
when using yolo_video.py Remember to modify class path or anchor path, with--classes class_file
and--anchors anchor_file
.
If you want to use original pretrained weights for YOLOv3:
1. wget https://pjreddie.com/media/files/darknet53.conv.74
2. rename it as darknet53.weights
3. python convert.py -w darknet53.cfg darknet53.weights model_data/darknet53_weights.h5
4. use model_data/darknet53_weights.h5 in train.py
-
The test environment is
- Python 3.5.2
- Keras 2.1.5
- tensorflow 1.6.0
-
Default anchors are used. If you use your own anchors, probably some changes are needed.
-
The inference result is not totally the same as Darknet but the difference is small.
-
The speed is slower than Darknet. Replacing PIL with opencv may help a little.
-
Always load pretrained weights and freeze layers in the first stage of training. Or try Darknet training. It's OK if there is a mismatch warning.
-
The training strategy is for reference only. Adjust it according to your dataset and your goal. And add further strategy if needed.
-
For speeding up the training process with frozen layers train_bottleneck.py can be used. It will compute the bottleneck features of the frozen model first and then only trains the last layers. This makes training on CPU possible in a reasonable time. See this for more information on bottleneck features.
-
In order to use our changes for vehicle detection, we changed the "tiny_yolo_body" function in "yolov3/model_its.py". As mentioned in paper, we removed " DarknetConv2D_BN_Leaky(1024, (3,3))" in line 105, "DarknetConv2D_BN_Leaky(128, (1,1))" (in line 112), and " DarknetConv2D_BN_Leaky(256, (3,3))," (in line 116) for Trial5 and also removed "BatchNormalization()" (in line 31) for Trial2 to Trial5.
@article{tajar2021lightweight,
title={A lightweight Tiny-YOLOv3 vehicle detection approach},
author={Tajar, Alireza Taheri and Ramazani, Abbas and Mansoorizadeh, Muharram},
journal={Journal of Real-Time Image Processing},
pages={1--13},
year={2021},
publisher={Springer}
}