本项目为基于Aidlux+r-retinanet+tflite，在小米平板5上实现热成像电力训练项目。通过r-retinanet对绝缘子等电力设施进行旋转目标检测。
首先，需要把老师提供的onnx导出为tflite模型，利用onnx2tflite.py进行转换。部分代码如下：
import osimport syssys.path.append("D:/wechat data/WeChat Files/wxid_ishz7g32wpon21/FileStorage/File/-04/onnx2tflite(1)/onnx2tflite")from converter import onnx_converterimport osos.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'def onnx2tflite(onnx_path):onnx_converter(onnx_model_path = onnx_path,need_simplify = False,output_path = os.path.dirname(onnx_path), target_formats = ['tflite'], # or ['keras'], ['keras', 'tflite']weight_quant = False,int8_model = False,int8_mean = None,int8_std = None,image_root = None)if __name__ == "__main__":onnx2tflite("./r-retinanet.onnx")
模型转换完成后，就可以在AidLux平台进行部署了。具体实现代码如下，这个代码是aidlux实现后置摄像头提取目标区域的。
if __name__=="__main__":cap = cvs.VideoCapture(0)frame_id = 0while True:frame = cap.read()if frame is None:continueframe_id += 1if frame_id % 3 != 0:continuetime0 = time.time()# 预处理im, im_scales = process_img(frame, NCHW=False, ToTensor=False) # im: NHWC#img = preprocess_img(frame, target_shape=(640, 640), div_num=255, means=None, stds=None)# img /= 255''' 定义输入输出shape '''in_shape = [1 * 640 * 800 * 3 * 4] # HWC, float32out_shape = [1 * 53325 * 8 * 4] # 8400: total cells, 52 = 48(num_classes) + 4(xywh), float32#out_shape = [1 * 55425 * 8 * 4] # 8400: total cells, 52 = 48(num_classes) + 4(xywh), float32''' AidLite初始化 '''aidlite = aidlite_gpu.aidlite()''' 加载R-RetinaNet模型 '''tflite_model = '/home/Lesson3_Training_and_Deploy/Lesson3_Training_and_Deploy/AidLux_Deploy/AidLux_Deploy/models/r-retinanet.tflite'res = aidlite.ANNModel(tflite_model, in_shape, out_shape, 4, -1) # Infer on -1: cpu, 0: gpu, 1: mixed, 2: dsp''' 设定输入输出 '''aidlite.setInput_Float32(im, 640, 800)''' 启动推理 '''aidlite.invoke()''' 捕获输出 '''preds = aidlite.getOutput_Float32(0)#preds = preds.reshape(1, 8, 53325)preds = preds.reshape(1, 8, (int)(preds.shape[0]/8))output = np.transpose(preds, (0, 2, 1))''' 创建Anchor '''im_anchor = np.transpose(im, (0, 3, 1, 2)).astype(np.float32)anchors_list = []anchor_generator = Anchors(ratios = np.array([0.2, 0.5, 1, 2, 5]))original_anchors = anchor_generator(im_anchor) # (bs, num_all_achors, 5)anchors_list.append(original_anchors)''' 解算输出 '''decode_output = decoder(im_anchor, anchors_list[-1], output[..., 5:8], output[..., 0:5], thresh=0.5, nms_thresh=0.2, test_conf=None)for i in range(len(decode_output)):print("dim({}), shape: {}".format(i, decode_output[i].shape))''' 重构输出 '''scores = decode_output[0].reshape(-1, 1)classes = decode_output[1].reshape(-1, 1)boxes = decode_output[2]boxes[:, :4] = boxes[:, :4] / im_scalesif boxes.shape[1] > 5: boxes[:, 5:9] = boxes[:, 5:9] / im_scalesdets = np.concatenate([classes, scores, boxes], axis=1)''' 过滤类别 '''keep = np.where(classes > 0)[0]dets = dets[keep, :]''' 转换坐标('xyxya'->'xyxyxyxy') '''res = sort_corners(rbox_2_quad(dets[:, 2:]))''' 评估绘图 '''for k in range(dets.shape[0]):cv2.line(frame, (int(res[k, 0]), int(res[k, 1])), (int(res[k, 2]), int(res[k, 3])), (0, 255, 0), 3)cv2.line(frame, (int(res[k, 2]), int(res[k, 3])), (int(res[k, 4]), int(res[k, 5])), (0, 255, 0), 3)cv2.line(frame, (int(res[k, 4]), int(res[k, 5])), (int(res[k, 6]), int(res[k, 7])), (0, 255, 0), 3)cv2.line(frame, (int(res[k, 6]), int(res[k, 7])), (int(res[k, 0]), int(res[k, 1])), (0, 255, 0), 3)cvs.imshow(frame)
最后，演示视频如下。
演示视频
如果觉得《利用AidLux实现热成像电力巡检项目-lesson3》对你有帮助，请点赞、收藏，并留下你的观点哦！
