Technologies used in - YOLOv3, YOLACT, Segmentation, Deep SORT, Kalman Filter and Smoother
This is a Driver Assistance System which can be used to predict the vehicles' trajectories. This system warns the driver if there is a danger ahead.
first of all install the requiremsnts to execute this program. For that run (pip install -r requirements.txt) in cmd
step 1 - Clone this repository and create three forlders inside the folder, (data/video). Name them as "cropped", "segment" and "raw1".
step 2 - download yolov3.weights file from here and put it inside the weights file. (we used the thir oned(YOLOv3-608))
| Model | Train | Test | mAP | FLOPS | Weights |
|---|---|---|---|---|---|
| YOLOv3-320 | COCO trainval | test-dev | 51.5 | 38.97 Bn | weights |
| YOLOv3-416 | COCO trainval | test-dev | 55.3 | 65.86 Bn | weights |
| YOLOv3-608 | COCO trainval | test-dev | 57.9 | 140.69 Bn | weights |
| YOLOv3-tiny | COCO trainval | test-dev | 33.1 | 5.56 Bn | weights |
step 3 - download yolact.pth file from here and put it inside the weights file. (we used the third one(yolact_base_54_800000.pth))
| Image Size | Backbone | FPS | mAP | Weights |
|---|---|---|---|---|
| 550 | Resnet50-FPN | 42.5 | 28.2 | yolact_resnet50_54_800000.pth |
| 550 | Darknet53-FPN | 40.0 | 28.7 | yolact_darknet53_54_800000.pth |
| 550 | Resnet101-FPN | 33.5 | 29.8 | yolact_base_54_800000.pth |
| 700 | Resnet101-FPN | 23.6 | 31.2 | yolact_im700_54_800000.pth |
step 4 - run the below command in the cmd to convert the yolov3.weights file into tensorflow supported weight file. "python convert.py --weights ./weights/yolov3.weights --output ./weights/yolov3.tf"
step 5 - put your video inside the folder, "data/video".
step 6 - Finally you need to run
python object_tracker.py --video ./(Location of the input video) --output ./data/video/result.avi
eg : python object_tracker.py --video ./data/video/001.mp4 --output ./data/video/result.avi
We get the markings of the most dangerous vehicles using ML as shown in the below image.
