updated

DetectorLoader.py

-import time
-import torch
-import numpy as np
-import torchvision.transforms as transforms
-
-from queue import Queue
-from threading import Thread
-
-from Detection.Models import Darknet
-from Detection.Utils import non_max_suppression, ResizePadding
-
-
-class TinyYOLOv3_onecls(object):
-    """Load trained Tiny-YOLOv3 one class (person) detection model.
-    Args:
-        input_size: (int) Size of input image must be divisible by 32. Default: 416,
-        config_file: (str) Path to Yolo model structure config file.,
-        weight_file: (str) Path to trained weights file.,
-        nms: (float) Non-Maximum Suppression overlap threshold.,
-        conf_thres: (float) Minimum Confidence threshold of predicted bboxs to cut off.,
-        device: (str) Device to load the model on 'cpu' or 'cuda'.
-    """
-    def __init__(self,
-                 input_size=416,
-                 config_file='mymodels/yolo-tiny-onecls/yolov3-tiny-onecls.cfg',
-                 weight_file='mymodels/yolo-tiny-onecls/best-model.pth',
-                 nms=0.2,
-                 conf_thres=0.45,
-                 device='cuda'):
-        self.input_size = input_size
-        self.model = Darknet(config_file).to(device)
-        self.model.load_state_dict(torch.load(weight_file,map_location=torch.device('cpu')))
-        self.model.eval()
-        self.device = device
-
-        self.nms = nms
-        self.conf_thres = conf_thres
-
-        self.resize_fn = ResizePadding(input_size, input_size)
-        self.transf_fn = transforms.ToTensor()
-
-    def detect(self, image, need_resize=True, expand_bb=5):
-        """Feed forward to the model.
-        Args:
-            image: (numpy array) Single RGB image to detect.,
-            need_resize: (bool) Resize to input_size before feed and will return bboxs
-                with scale to image original size.,
-            expand_bb: (int) Expand boundary of the boxs.
-        Returns:
-            (torch.float32) Of each detected object contain a
-                [top, left, bottom, right, bbox_score, class_score, class]
-            return `None` if no detected.
-        """
-        image_size = (self.input_size, self.input_size)
-        if need_resize:
-            image_size = image.shape[:2]
-            image = self.resize_fn(image)
-
-        image = self.transf_fn(image)[None, ...]
-        scf = torch.min(self.input_size / torch.FloatTensor([image_size]), 1)[0]
-
-        detected = self.model(image.to(self.device))
-        detected = non_max_suppression(detected, self.conf_thres, self.nms)[0]
-        if detected is not None:
-            detected[:, [0, 2]] -= (self.input_size - scf * image_size[1]) / 2
-            detected[:, [1, 3]] -= (self.input_size - scf * image_size[0]) / 2
-            detected[:, 0:4] /= scf
-
-            detected[:, 0:2] = np.maximum(0, detected[:, 0:2] - expand_bb)
-            detected[:, 2:4] = np.minimum(image_size[::-1], detected[:, 2:4] + expand_bb)
-
-        return detected
-
-
-class ThreadDetection(object):
-    def __init__(self,
-                 dataloader,
-                 model,
-                 queue_size=256):
-        self.model = model
-
-        self.dataloader = dataloader
-        self.stopped = False
-        self.Q = Queue(maxsize=queue_size)
-
-    def start(self):
-        t = Thread(target=self.update, args=(), daemon=True).start()
-        return self
-
-    def update(self):
-        while True:
-            if self.stopped:
-                return
-
-            images = self.dataloader.getitem()
-
-            outputs = self.model.detect(images)
-
-            if self.Q.full():
-                time.sleep(2)
-            self.Q.put((images, outputs))
-
-    def getitem(self):
-        return self.Q.get()
-
-    def stop(self):
-        self.stopped = True
-
-    def __len__(self):
-        return self.Q.qsize()
-
-
-
-
-
-
-

mymodels/yolo-tiny-onecls/yolov3-tiny-onecls.cfg

-[net]
-# Testing
-batch=1
-subdivisions=1
-# Training
-# batch=64
-# subdivisions=2
-width=416
-height=416
-channels=3
-momentum=0.9
-decay=0.0005
-angle=0
-saturation = 1.5
-exposure = 1.5
-hue=.1
-
-learning_rate=0.001
-burn_in=1000
-max_batches = 500200
-policy=steps
-steps=400000,450000
-scales=.1,.1
-
-# 0
-[convolutional]
-batch_normalize=1
-filters=16
-size=3
-stride=1
-pad=1
-activation=leaky
-
-# 1
-[maxpool]
-size=2
-stride=2
-
-# 2
-[convolutional]
-batch_normalize=1
-filters=32
-size=3
-stride=1
-pad=1
-activation=leaky
-
-# 3
-[maxpool]
-size=2
-stride=2
-
-# 4
-[convolutional]
-batch_normalize=1
-filters=64
-size=3
-stride=1
-pad=1
-activation=leaky
-
-# 5
-[maxpool]
-size=2
-stride=2
-
-# 6
-[convolutional]
-batch_normalize=1
-filters=128
-size=3
-stride=1
-pad=1
-activation=leaky
-
-# 7
-[maxpool]
-size=2
-stride=2
-
-# 8
-[convolutional]
-batch_normalize=1
-filters=256
-size=3
-stride=1
-pad=1
-activation=leaky
-
-# 9
-[maxpool]
-size=2
-stride=2
-
-# 10
-[convolutional]
-batch_normalize=1
-filters=512
-size=3
-stride=1
-pad=1
-activation=leaky
-
-# 11
-[maxpool]
-size=2
-stride=1
-
-# 12
-[convolutional]
-batch_normalize=1
-filters=1024
-size=3
-stride=1
-pad=1
-activation=leaky
-
-###########
-
-# 13
-[convolutional]
-batch_normalize=1
-filters=256
-size=1
-stride=1
-pad=1
-activation=leaky
-
-# 14
-[convolutional]
-batch_normalize=1
-filters=512
-size=3
-stride=1
-pad=1
-activation=leaky
-
-# 15
-[convolutional]
-size=1
-stride=1
-pad=1
-filters=18
-activation=linear
-
-
-
-# 16
-[yolo]
-mask = 3,4,5
-anchors = 10,14,  23,27,  37,58,  81,82,  135,169,  344,319
-classes=1
-num=6
-jitter=.3
-ignore_thresh = .7
-truth_thresh = 1
-random=1
-
-# 17
-[route]
-layers = -4
-
-# 18
-[convolutional]
-batch_normalize=1
-filters=128
-size=1
-stride=1
-pad=1
-activation=leaky
-
-# 19
-[upsample]
-stride=2
-
-# 20
-[route]
-layers = -1, 8
-
-# 21
-[convolutional]
-batch_normalize=1
-filters=256
-size=3
-stride=1
-pad=1
-activation=leaky
-
-# 22
-[convolutional]
-size=1
-stride=1
-pad=1
-filters=18
-activation=linear
-
-# 23
-[yolo]
-mask = 1,2,3
-anchors = 10,14,  23,27,  37,58,  81,82,  135,169,  344,319
-classes=1
-num=6
-jitter=.3
-ignore_thresh = .7
-truth_thresh = 1
-random=1