robot_ur3e_yolov5

git clone https://github.com/Andy-Leo10/robot_ur3e_yolov5.git

• robot_ur3e_yolov5
  • Motivation
  • Optional setup
    • tensorboard
    • pip in venv
  • trains
    • 1. e=10
    • 2. e=100
    • 3. e=300
    • 4. e=y
  • tests
    • 1. d=0
    • 2. d=0.86
    • 3. d=0.94
    • 4. e=y

---

Motivation

Motivation

When working on a real environment it is difficult to tune parameters in a program based on digital image processing. For that reason, a good approach is to use a CNN.

---

Optional setup

Optional setup

tensorboard

cd yolov5
tensorboard --logdir runs\train

pip in venv

.venv\Scripts\python.exe -m pip install <LIBRARY>

---

TRAINING

trains

1. e=10

python train.py --img 424 --batch 16 --epochs 10 --data customData.yaml --weights yolov5n.pt --cache

2. e=100

python train.py --img 424 --batch 16 --epochs 100 --data customData.yaml --weights yolov5n.pt --cache

3. e=300

python train.py --img 424 --batch 16 --epochs 300 --data customData.yaml --weights yolov5n.pt --cache

4. e=y

---

TESTING

tests

1. d=0

python detect.py --weights C:\Users\Leo\Downloads\MachineLearning\yolov5\runs\train\exp\weights\best.pt --img 424 --conf 0.25 --source C:\Users\Leo\Downloads\MachineLearning\photo_1718317391.jpg

2. d=0.86

python detect.py --weights C:\Users\Leo\Downloads\MachineLearning\yolov5\runs\train\exp1\weights\best.pt --img 424 --conf 0.25 --source C:\Users\Leo\Downloads\MachineLearning\photo_1718317391.jpg

3. d=0.94

python detect.py --weights C:\Users\Leo\Downloads\MachineLearning\yolov5\runs\train\exp2\weights\best.pt --img 424 --conf 0.25 --source C:\Users\Leo\Downloads\MachineLearning\photo_1718317391.jpg

4. e=y

---