Update README.md
Browse files
README.md
CHANGED
|
@@ -5,10 +5,16 @@ tags:
|
|
| 5 |
- computer vision
|
| 6 |
- darknet
|
| 7 |
- yolo
|
|
|
|
| 8 |
|
| 9 |
datasets:
|
| 10 |
- coco
|
| 11 |
- imagenette
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
---
|
| 13 |
|
| 14 |
# TensorFlow-yolov4-tflite
|
|
@@ -27,7 +33,6 @@ This yolov4 library uses Tensorflow 2.0 and is available on. this [Github](https
|
|
| 27 |
|
| 28 |
### What can you use yolov4-tflite for?
|
| 29 |
|
| 30 |
-
|
| 31 |
```bash
|
| 32 |
# Convert darknet weights to tensorflow
|
| 33 |
## yolov4
|
|
@@ -85,7 +90,7 @@ python save_model.py --weights ./data/yolov4.weights --output ./checkpoints/yolo
|
|
| 85 |
python convert_trt.py --weights ./checkpoints/yolov4.tf --quantize_mode float16 --output ./checkpoints/yolov4-trt-fp16-416
|
| 86 |
```
|
| 87 |
|
| 88 |
-
###
|
| 89 |
```bash
|
| 90 |
# run script in /script/get_coco_dataset_2017.sh to download COCO 2017 Dataset
|
| 91 |
# preprocess coco dataset
|
|
@@ -157,13 +162,6 @@ python benchmarks.py --size 416 --model yolov4 --weights ./data/yolov4.weights
|
|
| 157 |
| YoloV3 FPS | | | |
|
| 158 |
| YoloV4 FPS | | | |
|
| 159 |
|
| 160 |
-
### Traning your own model
|
| 161 |
-
```bash
|
| 162 |
-
# Prepare your dataset
|
| 163 |
-
# If you want to train from scratch:
|
| 164 |
-
In config.py set FISRT_STAGE_EPOCHS=0
|
| 165 |
-
# Run script:
|
| 166 |
-
python train.py
|
| 167 |
|
| 168 |
# Transfer learning:
|
| 169 |
python train.py --weights ./data/yolov4.weights
|
|
|
|
| 5 |
- computer vision
|
| 6 |
- darknet
|
| 7 |
- yolo
|
| 8 |
+
- tensorflow
|
| 9 |
|
| 10 |
datasets:
|
| 11 |
- coco
|
| 12 |
- imagenette
|
| 13 |
+
|
| 14 |
+
license: mit
|
| 15 |
+
|
| 16 |
+
thumbnail: https://github.com/hunglc007/tensorflow-yolov4-tflite
|
| 17 |
+
|
| 18 |
---
|
| 19 |
|
| 20 |
# TensorFlow-yolov4-tflite
|
|
|
|
| 33 |
|
| 34 |
### What can you use yolov4-tflite for?
|
| 35 |
|
|
|
|
| 36 |
```bash
|
| 37 |
# Convert darknet weights to tensorflow
|
| 38 |
## yolov4
|
|
|
|
| 90 |
python convert_trt.py --weights ./checkpoints/yolov4.tf --quantize_mode float16 --output ./checkpoints/yolov4-trt-fp16-416
|
| 91 |
```
|
| 92 |
|
| 93 |
+
### Evaluated on COCO 2017 Dataset
|
| 94 |
```bash
|
| 95 |
# run script in /script/get_coco_dataset_2017.sh to download COCO 2017 Dataset
|
| 96 |
# preprocess coco dataset
|
|
|
|
| 162 |
| YoloV3 FPS | | | |
|
| 163 |
| YoloV4 FPS | | | |
|
| 164 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 165 |
|
| 166 |
# Transfer learning:
|
| 167 |
python train.py --weights ./data/yolov4.weights
|