More Than YOLO

English | 中文

TensorFlow & Keras & Python

YOLOv3, YOLOv3-tiny, YOLOv4, YOLOv4-tiny

[Unofficial] YOLOv4-tiny, YOLOX

requirements: TensorFlow 2.1 (not test on 1.x), OpenCV, Numpy, PyYAML

---

News !

• Small batch size is used, because available GPU (8G) has small memory. Please use big batch size as possible.
• Online High Level augmentation will slow down training speed.
• When I tried to train yolov3 or yolov4, 'NaN' problem made me crazy.
• Support AccumOptimizer, Similar to 'subdivisions' in darknet.
• When I tried to train yolov3 or yolov4, I found that if I set weight decay to 5e-4，the result is unsatisfactory; if I set it to 0, everything is OK.
• Note that both yolov3-tiny and yolov4-tiny don't use anchor 0, so they use only anchors 1-7.[link]

---

This repository have done:

• Backbone for YOLO (YOLOv3, YOLOv3-tiny, YOLOv4, YOLOv4-tiny, Unofficial-YOLOv4-tiny)
• YOLOv3 Head
• Keras Callbacks for Online Evaluation
• Load Official Weight File
• Data Format Converter(COCO and Pascal VOC)
• K-Means for Anchors
• Fight with 'NaN'
• Train (Strategy and Model Config)
  • Define simple training in train.py
  • Use YAML as config file in cfgs
  • Cosine Annealing LR
  • Warm-up
  • AccumOptimizer
• Data Augmentation
  • Standard Method: Random Flip, Random Crop, Zoom, Random Grayscale, Random Distort, Rotate
  • Hight Level: Cut Mix, Mix Up, Mosaic （These Online Augmentations is Slow）
  • More, I can be so much more ...
• For Loss
  • Label Smoothing
  • Focal Loss
  • L2, D-IoU, G-IoU, C-IoU
  • ...

---

[toc]

0. Please Read Source Code for More Details

You can get official weight files from https://github.com/AlexeyAB/darknet/releases or https://pjreddie.com/darknet/yolo/.

1. Samples

1.1 Data File

We use a special data format like that,

path/to/image1 x1,y1,x2,y2,label x1,y1,x2,y2,label 
path/to/image2 x1,y1,x2,y2,label 
...

Convert your data format firstly. We present a script for Pascal VOC and a script for COCO.

More details and a simple dataset could be got from https://github.com/YunYang1994/yymnist.

1.2 Configure

# coco_yolov4.yaml
yolo:
  type: "yolov4"  # must be 'yolov3', 'yolov3_tiny', 'yolov4', 'yolov4_tiny' ‘unofficial_yolov4_tiny’ and 'yolox'.
  iou_threshold: 0.5
  score_threshold: 0.005
  max_boxes: 100
  strides: "32,16,8"
  anchors: "12,16 19,36 40,28 36,75 76,55 72,146 142,110 192,243 459,401"
  mask: "6,7,8 3,4,5 0,1,2"
  name_path: "./data/coco/coco.name"

train:
  label: "coco_yolov4" # any thing you like
  anno_path: "./data/coco/train2017.txt"
  image_size: "320,352,384,416,448,480,512,544,576,608"  # "416" for single mini batch size, "352,384,416,448,480" for Dynamic mini batch size.

  batch_size: 4
  init_weight_path: "./ckpts/yolov4.weights"
  save_weight_path: "./ckpts"

  loss_type: "CIoU+FL" # Must be "L2", "DIoU", "GIoU", "CIoU" or something like "L2+FL" for focal loss
  
  # turn on hight level data augmentation
  mosaic: false
  label_smoothing: false
  normal_method: true

  ignore_threshold: 0.7

test:
  anno_path: "./data/coco/val2017.txt"
  image_size: "608"
  batch_size: 1
  init_weight_path: "./ckpts/yolov4.weights"

1.3 K-Means

Edit the kmeans.py as you like

# kmeans.py
# Key Parameters
K = 6 # num of clusters
image_size = 416
dataset_path = './data/pascal_voc/train.txt'

1.4 Inference

A Simple Script for Video, Device or Image

Only support mp4, avi, device id, rtsp, png, jpg (Based on OpenCV)

python detector.py --config=./cfgs/coco_yolov4.yaml --media=./misc/street.mp4 --gpu=false

A simple demo for YOLOv4

from core.utils import decode_cfg, load_weights
from core.model.one_stage.yolov4 import YOLOv4
from core.image import draw_bboxes, preprocess_image, postprocess_image, read_image, Shader

import numpy as np
import cv2
import time

# read config
cfg = decode_cfg('cfgs/coco_yolov4.yaml')
names = cfg['yolo']['names']

model, eval_model = YOLOv4(cfg)
eval_model.summary()

# assign colors for difference labels
shader = Shader(cfg['yolo']['num_classes'])

# load weights
load_weights(model, cfg['test']['init_weight_path'])

img_raw = read_image('./misc/dog.jpg')
img = preprocess_image(img_raw, (512, 512))
imgs = img[np.newaxis, ...]

tic = time.time()
boxes, scores, classes, valid_detections = eval_model.predict(imgs)
toc = time.time()
print((toc - tic)*1000, 'ms')

# for single image, batch size is 1
valid_boxes = boxes[0][:valid_detections[0]]
valid_score = scores[0][:valid_detections[0]]
valid_cls = classes[0][:valid_detections[0]]

img, valid_boxes = postprocess_image(img, img_raw.shape[1::-1], valid_boxes)
img = draw_bboxes(img, valid_boxes, valid_score, valid_cls, names, shader)

cv2.imshow('img', img[..., ::-1])
cv2.imwrite('./misc/dog_v4.jpg', img)
cv2.waitKey()

2. Train

!!! Please Read the above guide (e.g. 1.1, 1.2).

python train.py --config=./cfgs/coco_yolov4.yaml

3. Experiment

3.1 Speed

i7-9700F+16GB

Model	416x416	512x512	608x608
YOLOv3	219 ms	320 ms	429 ms
YOLOv3-tiny	49 ms	63 ms	78 ms
YOLOv4	344 ms	490 ms	682 ms
YOLOv4-tiny	51 ms	66 ms	83 ms
Unofficial-YOLOv4-tiny	64 ms	86 ms	110 ms
YOLOX	67 ms	83 ms	104 ms

i7-9700F+16GB / RTX 2070S+8G

Model	416x416	512x512	608x608
YOLOv3	59 ms	66 ms	83 ms
YOLOv3-tiny	28 ms	30 ms	33 ms
YOLOv4	73 ms	74 ms	91 ms
YOLOv4-tiny	30 ms	32 ms	35 ms
Unofficial-YOLOv4-tiny	30 ms	31 ms	34 ms
YOLOX	42 ms	45 ms	50 ms

3.2 Logs

Augmentations

Name	Abbr
Standard Method	SM
Dynamic mini batch size	DM
Label Smoothing	LS
Focal Loss	FL
Mosaic	M
Warm-up LR	W
Cosine Annealing LR	CA

Standard Method Package includes Flip left and right, Crop and Zoom(jitter=0.3), Grayscale, Distort, Rotate(angle=7).

YOLOv3-tiny(Pretrained on COCO; Trained on VOC)

SM	DM	LS	FL	M	Loss	AP	AP@50	AP@75
✔					L2	26.6	61.8	17.2
✔	✔				L2	27.3	62.4	17.9
✔	✔	✔			L2	26.7	61.7	17.1
✔	✔				CIoU	30.9	64.2	25.0
✔	✔		✔		CIoU	32.3	65.7	27.6
✔	✔		✔	✔	CIoU

YOLOv3(TODO; Pretrained on COCO; Trained on VOC; only 15 epochs)

SM	DM	LS	FL	M	Loss	AP	AP@50	AP@75
✔	✔		✔		CIoU	46.5	80.0	49.0
✔	✔		✔	✔	CIoU

YOLOv4-tiny(TODO; Pretrained on COCO; Trained on VOC)

SM	DM	LS	FL	M	Loss	AP	AP@50	AP@75
✔	✔		✔		CIoU	35.1	70.2	30.0
✔	✔		✔	✔	CIoU

YOLOv4(TODO; Pretrained on COCO; Trained on VOC)

SM	DM	LS	FL	M	Loss	AP	AP@50	AP@75
✔	✔		✔		CIoU
✔	✔		✔	✔	CIoU

Unofficial-YOLOv4-tiny(TODO; Pretrained on COCO, part of YOLOv3-tiny weights; Trained on VOC)

SM	DM	LS	FL	M	Loss	AP	AP@50	AP@75
✔	✔		✔		CIoU	35.0	65.7	33.8
✔	✔		✔	✔	CIoU

YOLOX(TODO; Pretrained on COCO, part of YOLOv4-tiny weights; Trained on VOC)

SM	DM	LS	FL	M	Loss	AP	AP@50	AP@75
✔	✔		✔		CIoU	40.6	72.2	40.3
✔	✔		✔	✔	CIoU

3.3 Details

Tiny Version

Stage	Freeze Backbone	LR	Steps
0	Yes	1e-3 (w/ W)	4000
1	Yes	-	32*4000
2	No	1e-3 to 1e-6 (w/ CA)	48*4000

Common Version

Stage	Freeze Backbone	LR	Steps
0	Yes	1e-3 (w/ W)	4000
1	Yes	-	80*4000
2	No	1e-3 to 1e-6 (w/ CA)	120*4000

Training a complete network is time-consuming ...

4. Reference

• https://github.com/YunYang1994/TensorFlow2.0-Examples/tree/master/4-Object_Detection/YOLOV3
• https://github.com/hunglc007/tensorflow-yolov4-tflite
• https://github.com/experiencor/keras-yolo3

5. History

• Slim version: https://github.com/yuto3o/yolox/tree/slim
• Tensorflow2.0-YOLOv3: https://github.com/yuto3o/yolox/tree/yolov3-tf2