Skip to content

GitLab

Commit 00169466 authored by sunzhq2's avatar sunzhq2
Browse files

update yolo

parent 92c75df1
Hide whitespace changes
Inline Side-by-side
Showing with 37 additions and 316 deletions
yolo/yolov5_cmcc/=4.1.2 deleted 100644 → 0
View file @ 92c75df1
Looking in indexes: https://pypi.doubanio.com/simple
Requirement already satisfied: opencv-python in /root/miniconda3/envs/py310/lib/python3.10/site-packages (4.12.0.88)
Requirement already satisfied: numpy<2.3.0,>=2 in /root/miniconda3/envs/py310/lib/python3.10/site-packages (from opencv-python) (2.2.6)
yolo/yolov5_cmcc/inf.sh deleted 100644 → 0
View file @ 92c75df1
export HIP_PRINTF_DEBUG_FOR_FP64=0
nohup python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 7 2>&1 | tee result7.log &
yolo/yolov5_cmcc/migraphx.sh deleted 100644 → 0
View file @ 92c75df1
python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 0
python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 1
python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 2
python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 3
python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 4
python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 5
python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 6
python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 7
\ No newline at end of file
yolo/yolov5_cmcc/migraphx_eval.py.bak deleted 100644 → 0
View file @ 92c75df1
# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
"""
Validate a trained YOLOv5 model accuracy on a custom dataset
Usage:
$ python path/to/val.py --data coco128.yaml --weights yolov5s.pt --img 640
"""
import argparse
import json
import os
import sys
from pathlib import Path
from threading import Thread
import cv2
import numpy as np
import torch
from torchvision.transforms import Resize
import migraphx
from tqdm import tqdm
import glob
import time
FILE = Path(__file__).resolve()
ROOT = FILE.parents[0] # YOLOv5 root directory
if str(ROOT) not in sys.path:
sys.path.append(str(ROOT)) # add ROOT to PATH
ROOT = Path(os.path.relpath(ROOT, Path.cwd())) # relative
from models.experimental import attempt_load
from utils.datasets import create_dataloader
from utils.general import coco80_to_coco91_class, check_dataset, check_img_size, check_requirements, \
check_suffix, check_yaml, box_iou, non_max_suppression, scale_coords, xyxy2xywh, xywh2xyxy, \
increment_path, colorstr, print_args
from utils.metrics import ap_per_class, ConfusionMatrix
from utils.plots import output_to_target, plot_images, plot_val_study
from utils.torch_utils import select_device, time_sync
def process_batch(detections, labels, iouv):
"""
Return correct predictions matrix. Both sets of boxes are in (x1, y1, x2, y2) format.
Arguments:
detections (Array[N, 6]), x1, y1, x2, y2, conf, class
labels (Array[M, 5]), class, x1, y1, x2, y2
Returns:
correct (Array[N, 10]), for 10 IoU levels
"""
correct = torch.zeros(detections.shape[0], iouv.shape[0], dtype=torch.bool, device=iouv.device)
iou = box_iou(labels[:, 1:], detections[:, :4])
x = torch.where((iou >= iouv[0]) & (labels[:, 0:1] == detections[:, 5])) # IoU above threshold and classes match
if x[0].shape[0]:
matches = torch.cat((torch.stack(x, 1), iou[x[0], x[1]][:, None]), 1).cpu().numpy() # [label, detection, iou]
if x[0].shape[0] > 1:
matches = matches[matches[:, 2].argsort()[::-1]]
matches = matches[np.unique(matches[:, 1], return_index=True)[1]]
# matches = matches[matches[:, 2].argsort()[::-1]]
matches = matches[np.unique(matches[:, 0], return_index=True)[1]]
matches = torch.Tensor(matches).to(iouv.device)
correct[matches[:, 1].long()] = matches[:, 2:3] >= iouv
return correct
def migraphx_yolov(model, data_tensor):
# 将输入的tensor数据转换为numpy
data_numpy=data_tensor.detach().cpu().numpy()
device = torch.device("cuda")
# 注意：这里需要执行赋值操作，否则会造成migraphx中输入数据步长不对
img_data = np.zeros(data_numpy.shape).astype("float32")
for i in range(data_numpy.shape[0]):
img_data[i, :, :, :] = data_numpy[i, :, :, :]
# 执行推理
result = model.run({"images": img_data})
# 将结果转换为tensor
result0=torch.from_numpy(np.array(result[0], copy=False)).to(device)
return result0
def prepare_input(image):
input_img = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
input_img = cv2.resize(input_img, (640, 640))
input_img = input_img.transpose(2, 0, 1)
input_img = np.expand_dims(input_img, 0)
input_img = np.ascontiguousarray(input_img)
input_img = input_img.astype(np.float32)
input_img = input_img / 255
return input_img
def run(data,
weights=None, # model.pt path(s)
batch_size=1, # batch size
imgsz=640, # inference size (pixels)
conf_thres=0.001, # confidence threshold
iou_thres=0.65, # NMS IoU threshold
device='', # cuda device, i.e. 0 or 0,1,2,3 or cpu
single_cls=False, # treat as single-class dataset
save_hybrid=False, # save label+prediction hybrid results to *.txt
dataloader=None,
plots=False,
):
resultdir = os.path.join('results', device)
os.makedirs(resultdir, exist_ok=True)
# 初始化模型并选择相应的计算设备
device = select_device(device, batch_size=batch_size)
if os.path.isfile("/workspace/gaoruiqi_inference/cmcc-code/yolov5-6.0/yolov5m.mxr"):
model = migraphx.load("/workspace/gaoruiqi_inference/cmcc-code/yolov5-6.0/yolov5m.mxr")
else:
# 解析推理模型
max_input = {"images":[24,3,1024,1024]}
model = migraphx.parse_onnx(weights, map_input_dims=max_input)
# 获取模型输入/输出节点信息
inputs = model.get_inputs()
outputs = model.get_outputs()
# 获取模型的输入name
inputName = model.get_parameter_names()[0]
# 获取模型的输入尺寸
inputShape = inputs[inputName].lens()
inputHeight = int(inputShape[2])
inputWidth = int(inputShape[3])
migraphx.quantize_fp16(model)
# 模型编译
model.compile(t=migraphx.get_target("gpu") ,device_id=0)
gs = 32
imgsz = 640
# Data
data = check_dataset(data) # check
# Configure
is_coco = isinstance(data.get('val'), str) and data['val'].endswith('coco/val2017.txt') # COCO dataset
nc = int(data['nc']) # number of classes
iouv = torch.linspace(0.5, 0.95, 10).to(device) # iou vector for mAP@0.5:0.95
niou = iouv.numel()
# Dataloader
task = 'val' # path to val images
dataloader = create_dataloader(data[task], imgsz, batch_size, gs, single_cls, pad=0.5, rect=False,
prefix=colorstr(f'{task}: '))[0]
seen = 0
confusion_matrix = ConfusionMatrix(nc=nc)
class_map = coco80_to_coco91_class()
s = ('%20s' + '%11s' * 6) % ('Class', 'Images', 'Labels', 'P', 'R', 'mAP@.5', 'mAP@.5:.95')
dt, p, r, f1, mp, mr, map50, map = [0.0, 0.0, 0.0], 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0
jdict, stats, ap, ap_class = [], [], [], []
# 数据预处理
i = 0
for batch_i, (img, targets, paths, shapes) in enumerate(tqdm(dataloader, desc=s)):
img = img.to(device, non_blocking=True)
img = img.half()
img /= 255.0 # 0 - 255 to 0.0 - 1.0
targets = targets.to(device)
nb, _, height, width = img.shape # batch size, channels, height, width
# Run model
out = migraphx_yolov(model, img)
# save to file
out.cpu().numpy().tofile(f'{resultdir}/{i}_0.bin')
i += 1
# Run NMS
targets[:, 2:] *= torch.Tensor([width, height, width, height]).to(device) # to pixels
lb = [targets[targets[:, 0] == i, 1:] for i in range(nb)] if save_hybrid else [] # for autolabelling
out = non_max_suppression(out, conf_thres, iou_thres, labels=lb, multi_label=True, agnostic=single_cls)
# Statistics per image
for si, pred in enumerate(out):
labels = targets[targets[:, 0] == si, 1:]
nl = len(labels)
tcls = labels[:, 0].tolist() if nl else [] # target class
path, shape = Path(paths[si]), shapes[si][0]
seen += 1
if len(pred) == 0:
if nl:
stats.append((torch.zeros(0, niou, dtype=torch.bool), torch.Tensor(), torch.Tensor(), tcls))
continue
# Predictions
if single_cls:
pred[:, 5] = 0
predn = pred.clone()
scale_coords(img[si].shape[1:], predn[:, :4], shape, shapes[si][1]) # native-space pred
# Evaluate
if nl:
tbox = xywh2xyxy(labels[:, 1:5]) # target boxes
scale_coords(img[si].shape[1:], tbox, shape, shapes[si][1]) # native-space labels
labelsn = torch.cat((labels[:, 0:1], tbox), 1) # native-space labels
correct = process_batch(predn, labelsn, iouv)
if plots:
confusion_matrix.process_batch(predn, labelsn)
else:
correct = torch.zeros(pred.shape[0], niou, dtype=torch.bool)
stats.append((correct.cpu(), pred[:, 4].cpu(), pred[:, 5].cpu(), tcls)) # (correct, conf, pcls, tcls)
# 计算统计数据
stats = [np.concatenate(x, 0) for x in zip(*stats)] # to numpy
if len(stats) and stats[0].any():
p, r, ap, f1, ap_class = ap_per_class(*stats, plot=False)
ap50, ap = ap[:, 0], ap.mean(1) # AP@0.5, AP@0.5:0.95
mp, mr, map50, map = p.mean(), r.mean(), ap50.mean(), ap.mean()
nt = np.bincount(stats[3].astype(np.int64), minlength=nc) # number of targets per class
else:
nt = torch.zeros(1)
# Print results
print('map50:', map50)
print('map50-95:', map)
def parse_opt():
parser = argparse.ArgumentParser()
parser.add_argument('--data', type=str, default=ROOT / 'data/coco128.yaml', help='dataset.yaml path')
parser.add_argument('--weights', type=str, default='', help='model.onnx path(s)')
parser.add_argument('--batch-size', type=int, default=32, help='batch size')
parser.add_argument('--imgsz', '--img', '--img-size', type=int, default=640, help='inference size (pixels)')
parser.add_argument('--conf-thres', type=float, default=0.001, help='confidence threshold')
parser.add_argument('--iou-thres', type=float, default=0.6, help='NMS IoU threshold')
parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
opt = parser.parse_args()
opt.data = check_yaml(opt.data) # check YAML
print_args(FILE.stem, opt)
return opt
def main(opt):
# 检测requirements文件中需要的包是否安装好了
check_requirements(exclude=('tensorboard', 'thop'))
run(**vars(opt))
if __name__ == "__main__":
opt = parse_opt()
main(opt)
yolo/yolov5_cmcc/migraphx_eval.py → yolo/yolov5_cmcc/tools/migraphx_eval.py
View file @ 00169466
......@@ -98,23 +98,6 @@ def process_batch(detections, labels, iouv):
correct[matches[:, 1].long()] = matches[:, 2:3] >= iouv
return correct
def migraphx_yolov(model, data_tensor):
# 将输入的tensor数据转换为numpy
data_numpy=data_tensor.detach().cpu().numpy()
device = torch.device("cuda")
# 注意：这里需要执行赋值操作，否则会造成migraphx中输入数据步长不对
img_data = np.zeros(data_numpy.shape).astype("float32")
for i in range(data_numpy.shape[0]):
img_data[i, :, :, :] = data_numpy[i, :, :, :]
# 执行推理
result = model.run({"images": img_data})
# 将结果转换为tensor
result0=torch.from_numpy(np.array(result[0], copy=False)).to(device)
return result0
def prepare_input(image):
input_img = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
......@@ -147,33 +130,21 @@ def run(data,
# 初始化模型并选择相应的计算设备
device = select_device(device, batch_size=batch_size)
if os.path.isfile("/home/sunzhq/workspace/yidong/yolo/yolov5_cmcc/yolov5m_fp16.mxr"):
model = migraphx.load("/home/sunzhq/workspace/yidong/yolo/yolov5_cmcc/yolov5m_fp16.mxr")
else:
if weights.split(".")[-1] == "mxr":
model = migraphx.load(weights)
inputName=list(model.get_inputs().keys())[0]
elif weights.split(".")[-1] == "onnx":
# 解析推理模型
max_input = {"images":[24,3,640,640]}
model = migraphx.parse_onnx(weights, map_input_dims=max_input)
# 获取模型输入/输出节点信息
inputs = model.get_inputs()
outputs = model.get_outputs()
# 获取模型的输入name
inputName = model.get_parameter_names()[0]
# 获取模型的输入尺寸
inputShape = inputs[inputName].lens()
inputHeight = int(inputShape[2])
inputWidth = int(inputShape[3])
migraphx.quantize_fp16(model)
# 模型编译
model.compile(t=migraphx.get_target("gpu"), offlod_copy=False, device_id=0)
inputName=list(model.get_inputs().keys())[0]
modelData=AllocateOutputMemory(model)
else:
print("请输出正确的模型路径")
modelData=AllocateOutputMemory(model)
gs = 32
imgsz = 640
......@@ -205,21 +176,19 @@ def run(data,
total_infer_times = []
total_start = time.time()
all_images = []
for batch_i, (img, targets, paths, shapes) in enumerate(tqdm(dataloader, desc=s)):
for img, _, _, _ in dataloader:
img = img.float() / 255.0
img_np=img.numpy()
all_images.append(img_np.astype(np.float32))
# warm up
modelData[inputName] = migraphx.to_gpu(migraphx.argument(all_images[0]))
model.run(modelData)
for batch_i, (_, targets, paths, shapes) in enumerate(tqdm(dataloader, desc=s)):
# modelData[inputName] = migraphx.to_gpu(migraphx.argument(img_data))
# preds_dcu = model.run(modelData)
# img = img.to(device, non_blocking=True)
img = all_images[batch_i]
modelData[inputName] = migraphx.to_gpu(migraphx.argument(img))
# img = img.half()
# img /= 255.0 # 0 - 255 to 0.0 - 1.0
targets = targets.to(device)
nb, _, height, width = img.shape # batch size, channels, height, width
......@@ -227,7 +196,6 @@ def run(data,
break
start = time.time()
# Run model
# out = migraphx_yolov(model, img)
out = model.run(modelData)
infer_times.append(time.time() - start)
total_infer_times.append(time.time() - total_start)
......@@ -255,7 +223,7 @@ def run(data,
save_coco_json(pred, pred_results, image_id, coco80_to_coco91_class())
total_start = time.time()
pred_json_file = f"yolov5m_predictions{this_file_device}.json"
pred_json_file = f"./results/yolov5m_predictions{this_file_device}.json"
with open(pred_json_file, 'w') as f:
json.dump(pred_results, f)
......@@ -278,8 +246,8 @@ def run(data,
def parse_opt():
parser = argparse.ArgumentParser()
parser.add_argument('--data', type=str, default=ROOT / 'data/coco128.yaml', help='dataset.yaml path')
parser.add_argument('--weights', type=str, default='', help='model.onnx path(s)')
parser.add_argument('--data', type=str, default=ROOT / '../data/coco128.yaml', help='dataset.yaml path')
parser.add_argument('--weights', type=str, default=' ', help='model.onnx path(s)')
parser.add_argument('--batch-size', type=int, default=32, help='batch size')
parser.add_argument('--imgsz', '--img', '--img-size', type=int, default=640, help='inference size (pixels)')
parser.add_argument('--conf-thres', type=float, default=0.001, help='confidence threshold')
......
yolo/yolov5_cmcc/onnx_inference.sh → yolo/yolov5_cmcc/tools/onnx_inference.sh
View file @ 00169466
......@@ -2,15 +2,15 @@ export PYTHONPATH=/opt/dtk/lib:$PYTHONPATH
export HIP_PRINTF_DEBUG_FOR_FP64=0
export HIP_VISIBLE_DEVICES=0
# python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 0
python ./tools/migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight /models/yolov5m_fp16.mxr --device 0
nohup numactl -N 0 -m 0 python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 0 2>&1 | tee result_0.log &
export HIP_VISIBLE_DEVICES=1
nohup numactl -N 1 -m 1 python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 1 2>&1 | tee result_1.log &
export HIP_VISIBLE_DEVICES=2
nohup numactl -N 2 -m 2 python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 2 2>&1 | tee result_2.log &
export HIP_VISIBLE_DEVICES=3
nohup numactl -N 3 -m 3 python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 3 2>&1 | tee result_3.log &
# nohup numactl -N 0 -m 0 python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 0 2>&1 | tee result_0.log &
# export HIP_VISIBLE_DEVICES=1
# nohup numactl -N 1 -m 1 python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 1 2>&1 | tee result_1.log &
# export HIP_VISIBLE_DEVICES=2
# nohup numactl -N 2 -m 2 python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 2 2>&1 | tee result_2.log &
# export HIP_VISIBLE_DEVICES=3
# nohup numactl -N 3 -m 3 python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 3 2>&1 | tee result_3.log &
# nohup python migraphx_eval.py --img 640 --batch-size 24 --data coco.yaml --weight ./yolov5m.onnx --device 4 2>&1 | tee result_4.log &
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment