Resume hyp tune

.github/workflows/ci-testing.yml

@@ -12,63 +12,22 @@ on:
     - cron: '0 0 * * *'  # runs at 00:00 UTC every day
 
 jobs:
-  Benchmarks:
-    runs-on: ${{ matrix.os }}
-    strategy:
-      fail-fast: false
-      matrix:
-        os: [ ubuntu-latest ]
-        python-version: [ '3.10' ]  # requires python<=3.10
-        model: [ yolov5n ]
-    steps:
-      - uses: actions/checkout@v3
-      - uses: actions/setup-python@v4
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip' # caching pip dependencies
-      - name: Install requirements
-        run: |
-          python -m pip install --upgrade pip wheel
-          pip install -r requirements.txt coremltools openvino-dev tensorflow-cpu --extra-index-url https://download.pytorch.org/whl/cpu
-          python --version
-          pip --version
-          pip list
-      - name: Benchmark DetectionModel
-        run: |
-          python benchmarks.py --data coco128.yaml --weights ${{ matrix.model }}.pt --img 320 --hard-fail 0.29
-      - name: Benchmark SegmentationModel
-        run: |
-          python benchmarks.py --data coco128-seg.yaml --weights ${{ matrix.model }}-seg.pt --img 320 --hard-fail 0.22
-      - name: Test predictions
-        run: |
-          python export.py --weights ${{ matrix.model }}-cls.pt --include onnx --img 224
-          python detect.py --weights ${{ matrix.model }}.onnx --img 320
-          python segment/predict.py --weights ${{ matrix.model }}-seg.onnx --img 320
-          python classify/predict.py --weights ${{ matrix.model }}-cls.onnx --img 224
-
   Tests:
     timeout-minutes: 60
     runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
-        os: [ ubuntu-latest, windows-latest ]  # macos-latest bug https://github.com/ultralytics/yolov5/pull/9049
+        os: [ ubuntu-latest ]  # macos-latest bug https://github.com/ultralytics/yolov5/pull/9049
         python-version: [ '3.10' ]
         model: [ yolov5n ]
         include:
-          - os: ubuntu-latest
-            python-version: '3.7'  # '3.6.8' min
-            model: yolov5n
           - os: ubuntu-latest
             python-version: '3.8'
             model: yolov5n
           - os: ubuntu-latest
             python-version: '3.9'
             model: yolov5n
-          - os: ubuntu-latest
-            python-version: '3.8'  # torch 1.7.0 requires python >=3.6, <=3.8
-            model: yolov5n
-            torch: '1.7.0'  # min torch version CI https://pypi.org/project/torchvision/
     steps:
       - uses: actions/checkout@v3
       - uses: actions/setup-python@v4
@@ -97,57 +56,17 @@ jobs:
           pip --version
           pip list
       - name: Test detection
-        shell: bash  # for Windows compatibility
+        shell: bash
         run: |
           # export PYTHONPATH="$PWD"  # to run '$ python *.py' files in subdirectories
           m=${{ matrix.model }}  # official weights
           b=runs/train/exp/weights/best  # best.pt checkpoint
           python train.py --imgsz 64 --batch 32 --weights $m.pt --cfg $m.yaml --epochs 1 --device cpu  # train
-          for d in cpu; do  # devices
-            for w in $m $b; do  # weights
-              python val.py --imgsz 64 --batch 32 --weights $w.pt --device $d  # val
-              python detect.py --imgsz 64 --weights $w.pt --device $d  # detect
-            done
-          done
-          python hubconf.py --model $m  # hub
-          # python models/tf.py --weights $m.pt  # build TF model
+          #for d in cpu; do  # devices
+          #  for w in $m $b; do  # weights
+          #    python val.py --imgsz 64 --batch 32 --weights $w.pt --device $d  # val
+          #    python detect.py --imgsz 64 --weights $w.pt --device $d  # detect
+          #  done
+          #done
           python models/yolo.py --cfg $m.yaml  # build PyTorch model
-          python export.py --weights $m.pt --img 64 --include torchscript  # export
-          python - <<EOF
-          import torch
-          im = torch.zeros([1, 3, 64, 64])
-          for path in '$m', '$b':
-              model = torch.hub.load('.', 'custom', path=path, source='local')
-              print(model('data/images/bus.jpg'))
-              model(im)  # warmup, build grids for trace
-              torch.jit.trace(model, [im])
-          EOF
-      - name: Test segmentation
-        shell: bash  # for Windows compatibility
-        run: |
-          m=${{ matrix.model }}-seg  # official weights
-          b=runs/train-seg/exp/weights/best  # best.pt checkpoint
-          python segment/train.py --imgsz 64 --batch 32 --weights $m.pt --cfg $m.yaml --epochs 1 --device cpu  # train
-          python segment/train.py --imgsz 64 --batch 32 --weights '' --cfg $m.yaml --epochs 1 --device cpu  # train
-          for d in cpu; do  # devices
-            for w in $m $b; do  # weights
-              python segment/val.py --imgsz 64 --batch 32 --weights $w.pt --device $d  # val
-              python segment/predict.py --imgsz 64 --weights $w.pt --device $d  # predict
-              python export.py --weights $w.pt --img 64 --include torchscript --device $d  # export
-            done
-          done
-      - name: Test classification
-        shell: bash  # for Windows compatibility
-        run: |
-          m=${{ matrix.model }}-cls.pt  # official weights
-          b=runs/train-cls/exp/weights/best.pt  # best.pt checkpoint
-          python classify/train.py --imgsz 32 --model $m --data mnist160 --epochs 1  # train
-          python classify/val.py --imgsz 32 --weights $b --data ../datasets/mnist160  # val
-          python classify/predict.py --imgsz 32 --weights $b --source ../datasets/mnist160/test/7/60.png  # predict
-          python classify/predict.py --imgsz 32 --weights $m --source data/images/bus.jpg  # predict
-          python export.py --weights $b --img 64 --include torchscript  # export
-          python - <<EOF
-          import torch
-          for path in '$m', '$b':
-              model = torch.hub.load('.', 'custom', path=path, source='local')
-          EOF
+          python export.py --weights $m.pt --img 64 --rknpu --int8 --data data/rknpu-qnt.txt # export

.gitignore

@@ -255,3 +255,7 @@ com_crashlytics_export_strings.xml
 crashlytics.properties
 crashlytics-build.properties
 fabric.properties
+
+# RKNN files
+*.rknn
+RK_anchors.txt

README.md

@@ -1,3 +1,11 @@
+#### Get model optimized for RKNN
+Exporting detect/segment model with optimization for RKNN, please refer here [README_rkopt.md](./README_rkopt.md)
+
+导出适配 RKNPU 的检测/分割模型, 请参考 [README_rkopt.md](./README_rkopt.md) 的说明
+
+---
+<br>
+
 <div align="center">
   <p>
     <a align="center" href="https://ultralytics.com/yolov5" target="_blank">

README_rkopt.md

@@ -0,0 +1,43 @@
+# YOLOv5 - RKNN optimize
+
+## Source
+
+  Base on https://github.com/ultralytics/yolov5 (v7.0) with commit id as 915bbf294bb74c859f0b41f1c23bc395014ea679
+
+
+
+## What different
+
+With inference result values unchanged, the following optimizations were applied:
+
+- Optimize focus/SPPF block, getting better performance with same result
+- Change output node, remove post_process from the model. (post process block in model is unfriendly for quantization)
+
+
+
+With inference result got changed, the following optimization was applied:
+
+- Using ReLU as activation layer instead of SiLU（Only valid when training new model）
+
+
+
+## How to use
+
+```
+# for detection model
+python export.py --rknpu --weight yolov5s.pt
+
+# for segmentation model
+python export.py --rknpu --weight yolov5s-seg.pt
+```
+
+- 'yolov5s.pt'/ 'yolov5s-seg.pt' could be replaced with your model path
+- A file name "RK_anchors.txt" would be generated and it would be used for the post_process stage. 
+- **NOTICE: Please call with --rknpu, do not changing the default rknpu value in export.py.** 
+
+
+
+## Deploy demo
+
+Please refer https://github.com/airockchip/rknn_model_zoo
+

classify/predict.py

@@ -46,7 +46,7 @@
 from models.common import DetectMultiBackend
 from utils.augmentations import classify_transforms
 from utils.dataloaders import IMG_FORMATS, VID_FORMATS, LoadImages, LoadScreenshots, LoadStreams
-from utils.general import (LOGGER, Profile, check_file, check_img_size, check_imshow, check_requirements, colorstr, cv2,
+from utils.general import (LOGGER, Profile, check_file, check_img_size, check_imshow,  colorstr, cv2,
                            increment_path, print_args, strip_optimizer)
 from utils.plots import Annotator
 from utils.torch_utils import select_device, smart_inference_mode
@@ -217,7 +217,6 @@ def parse_opt():
 
 
 def main(opt):
-    check_requirements(exclude=('tensorboard', 'thop'))
     run(**vars(opt))
 
 

classify/train.py

@@ -40,8 +40,8 @@
 from models.experimental import attempt_load
 from models.yolo import ClassificationModel, DetectionModel
 from utils.dataloaders import create_classification_dataloader
-from utils.general import (DATASETS_DIR, LOGGER, TQDM_BAR_FORMAT, WorkingDirectory, check_git_info, check_git_status,
-                           check_requirements, colorstr, download, increment_path, init_seeds, print_args, yaml_save)
+from utils.general import (DATASETS_DIR, LOGGER, TQDM_BAR_FORMAT, WorkingDirectory, check_git_info, 
+                           colorstr, download, increment_path, init_seeds, print_args, yaml_save)
 from utils.loggers import GenericLogger
 from utils.plots import imshow_cls
 from utils.torch_utils import (ModelEMA, de_parallel, model_info, reshape_classifier_output, select_device, smart_DDP,
@@ -299,8 +299,6 @@ def main(opt):
     # Checks
     if RANK in {-1, 0}:
         print_args(vars(opt))
-        check_git_status()
-        check_requirements()
 
     # DDP mode
     device = select_device(opt.device, batch_size=opt.batch_size)

classify/val.py

@@ -36,7 +36,7 @@
 
 from models.common import DetectMultiBackend
 from utils.dataloaders import create_classification_dataloader
-from utils.general import (LOGGER, TQDM_BAR_FORMAT, Profile, check_img_size, check_requirements, colorstr,
+from utils.general import (LOGGER, TQDM_BAR_FORMAT, Profile, check_img_size, colorstr,
                            increment_path, print_args)
 from utils.torch_utils import select_device, smart_inference_mode
 
@@ -161,7 +161,6 @@ def parse_opt():
 
 
 def main(opt):
-    check_requirements(exclude=('tensorboard', 'thop'))
     run(**vars(opt))
 
 

data/Objects365.yaml

@@ -386,9 +386,8 @@ names:
 download: |
   from tqdm import tqdm
 
-  from utils.general import Path, check_requirements, download, np, xyxy2xywhn
+  from utils.general import Path, download, np, xyxy2xywhn
 
-  check_requirements(('pycocotools>=2.0',))
   from pycocotools.coco import COCO
 
   # Make Directories

data/rknpu-qnt.txt

@@ -0,0 +1,2 @@
+images/bus.jpg
+images/zidane.jpg

detect.py

@@ -44,7 +44,7 @@
 
 from models.common import DetectMultiBackend
 from utils.dataloaders import IMG_FORMATS, VID_FORMATS, LoadImages, LoadScreenshots, LoadStreams
-from utils.general import (LOGGER, Profile, check_file, check_img_size, check_imshow, check_requirements, colorstr, cv2,
+from utils.general import (LOGGER, Profile, check_file, check_img_size, check_imshow, colorstr, cv2,
                            increment_path, non_max_suppression, print_args, scale_boxes, strip_optimizer, xyxy2xywh)
 from utils.plots import Annotator, colors, save_one_box
 from utils.torch_utils import select_device, smart_inference_mode
@@ -252,7 +252,6 @@ def parse_opt():
 
 
 def main(opt):
-    check_requirements(exclude=('tensorboard', 'thop'))
     run(**vars(opt))