Upload 406 files

.gitattributes

@@ -52,3 +52,104 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.jpg filter=lfs diff=lfs merge=lfs -text
 *.jpeg filter=lfs diff=lfs merge=lfs -text
 *.webp filter=lfs diff=lfs merge=lfs -text
+bdd100k/bdd100k.torrent filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc30fc-e7726578.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc30fc-eb673c5a.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc30fc-fd79926f.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc9045-1b8282ba.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc9045-581f64de.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc9045-5a50690f.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc9045-b3349548.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc9045-c6dc9529.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc9045-cd422b81.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabc9045-d91ecb66.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabddb96-ca0ac856.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabe1040-5f02711e.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabe1040-c59cb390.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabea010-6882cf41.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabf7be1-36a39a28.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabf7be1-f1a7e00d.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cabf9f3c-d58a6760.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-0396e053.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-0eb1c8bf.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-15b814db.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-196cd6f8.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-76e4c968.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-951977c8.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-ba37148a.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-bc0b048a.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-e969f06a.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac07407-fe32e494.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac32276-2ea56b83.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac32276-a70feba7.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac32276-cf233a28.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac47e88-3227e13a.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac6a8d8-29780d2d.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cac75967-8a9c164b.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cace2484-cd478fab.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cacfaddc-ce8105be.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad02f4a-dd2c4b41.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad03998-075bb24b.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad03998-b2cc3e61.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad03998-e6cc2ae6.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad17389-cb0208f5.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad17389-f72e5365.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad180c4-2edc9315.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad180c4-553ceeb1.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad180c4-d05b49f1.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad4a646-68a8d136.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad54cf7-6779254f.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cad7fdff-d9946f73.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadb239d-0776327a.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadb239d-499b1478.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadb239d-a768df31.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadb5cf7-31d9a0d7.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadbd8df-9efc61fd.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadc60a4-0834935f.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadd00a0-4251d603.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadd00a0-efef6501.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cade2084-5714bb13.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cade2084-71df1a4e.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cade2084-8fb02395.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cade2aed-db2e9c98.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadf297e-457ef8ae.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadf77c3-2857db38.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadfb2fe-5dd5f809.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadfb2fe-708d6e3b.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cadfb2fe-f43878a3.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae23877-a1826db4.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae28701-5286ae2f.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae29b60-b9d5bdb6.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae39588-58dd3092.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae39588-8ea784d1.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae49ab7-c87d52c4.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae4f10f-54b690b0.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae4f10f-e5906613.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae6c88d-0db9c3ab.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae73656-aeb2944f.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae77018-7bc11357.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae842b9-82d52302.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae881fa-02da1ae2.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae8e3d7-13803eef.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae913d9-6df0f74b.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/cae913d9-e07a8369.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caea91fd-e457bf7b.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caeb0fbb-410c10ca.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caeb782d-4a20b7c4.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caec69a1-75429ccd.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caee33ed-136d6b7c.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caee33ed-bb44010a.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf380ad-52f6a603.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf380ad-83f39185.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf380ad-ee9a7699.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf5502b-8233af1b.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf56f7a-1407df75.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf6713b-1823e54e.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf6d8fe-7650f811.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf73e65-16760e37.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf73e65-99596d6e.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf8496d-6ce3dbda.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf91e5e-fbd7f315.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf92153-8fdace82.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf92153-cfcdf92a.mov filter=lfs diff=lfs merge=lfs -text
+bdd100k/videos/test/caf9b4a2-dbf2a8a7.mov filter=lfs diff=lfs merge=lfs -text

bdd100k/K-means.py

@@ -0,0 +1,129 @@
+# Author:LiPu
+import numpy as np
+import pandas as pd
+import os
+
+
+class YOLO_Kmeans:
+
+    def __init__(self, cluster_number, data_file, anchors_file):
+        self.cluster_number = cluster_number  # 6 or 9
+        self.data_file = data_file
+        self.anchors_file = anchors_file
+
+    def iou(self, boxes, clusters):  # 1 box -> k clusters, boxes代表实际的每个box的宽和高， clusters为聚类中心
+        n = boxes.shape[0]
+        k = self.cluster_number
+
+        box_area = boxes[:, 0] * boxes[:, 1]  # box的面积长x宽
+
+        box_area = box_area.repeat(k)  # 将每个box的面积重复6次，构成一个1行6列的数组
+
+        box_area = np.reshape(box_area, (n, k))
+
+        cluster_area = clusters[:, 0] * clusters[:, 1]  # 计算每个聚类中心的面积
+        cluster_area = np.tile(cluster_area, [1, n])
+        cluster_area = np.reshape(cluster_area, (n, k))  # 将聚类中心构成与输入boxes具有同样维度的矩阵，可以避免循环运算
+
+        box_w_matrix = np.reshape(boxes[:, 0].repeat(k), (n, k))
+        cluster_w_matrix = np.reshape(np.tile(clusters[:, 0], (1, n)), (n, k))
+        min_w_matrix = np.minimum(cluster_w_matrix, box_w_matrix)
+
+        box_h_matrix = np.reshape(boxes[:, 1].repeat(k), (n, k))
+        cluster_h_matrix = np.reshape(np.tile(clusters[:, 1], (1, n)), (n, k))
+        min_h_matrix = np.minimum(cluster_h_matrix, box_h_matrix)
+        inter_area = np.multiply(min_w_matrix, min_h_matrix)
+
+        result = inter_area / (box_area + cluster_area - inter_area)
+        return result
+
+    def avg_iou(self, boxes, clusters):
+        accuracy = np.mean([np.max(self.iou(boxes, clusters), axis=1)])
+        return accuracy
+
+    def kmeans(self, boxes, k, dist=np.median):
+        box_number = boxes.shape[0]
+        distances = np.empty((box_number, k))
+        last_nearest = np.zeros((box_number,))
+        np.random.seed()
+        clusters = boxes[np.random.choice(
+            box_number, k, replace=False)]  # init k clusters从原来的众多box中随机选取6个box
+        while True:
+
+            distances = 1 - self.iou(boxes, clusters)
+
+            current_nearest = np.argmin(distances, axis=1)
+            if (last_nearest == current_nearest).all():
+                break  # clusters won't change
+            for cluster in range(k):
+                clusters[cluster] = dist(  # update clusters
+                    boxes[current_nearest == cluster], axis=0)
+
+            last_nearest = current_nearest
+
+        return clusters
+
+    def result2txt(self, data):  # 将data保存进txt文档中，覆盖方式
+        f = open(self.anchors_file, 'w')
+        row = np.shape(data)[0]
+        for i in range(row):
+            if i == 0:
+                x_y = "%d,%d" % (data[i][0], data[i][1])  # 若data只有1行，直接两个数据用逗号隔开保存
+            else:
+                x_y = ", %d,%d" % (data[i][0], data[i][1])  # 若data有多行，就将每行数据以空格隔开
+            f.write(x_y)
+        f.close()
+
+    def txt2boxes(self):
+        test = os.listdir(self.data_file + 'test/')
+        train = os.listdir(self.data_file + 'train/')
+        dataSet = []
+        for i in train:
+            f = open(self.data_file + 'train/' + i, 'r')
+            for line in f:
+                infos = line.split(" ")
+                width = int(float(infos[3]) * 416)
+                height = int(float(infos[4]) * 416)
+                dataSet.append([width, height])
+            f.close()
+
+        for i in test:
+            f = open(self.data_file + 'test/' + i, 'r')
+            for line in f:
+                infos = line.split(" ")
+                width = int(float(infos[3]) * 416)
+                height = int(float(infos[4]) * 416)
+                dataSet.append([width, height])
+            f.close()
+        result = np.array(dataSet)
+        return result
+
+    def csv2boxes(self):
+        dataSet = []
+        train_data = pd.read_csv(self.data_file, header=None)
+        for value_i in train_data.values:
+            end_num = len(value_i)
+            for i in range(1, end_num, 5):
+                width = value_i[i + 2] - value_i[i]
+                height = value_i[i + 3] - value_i[i + 1]
+                dataSet.append([width, height])  ##获取所有box的宽和高
+        result = np.array(dataSet)
+        return result
+
+    def txt2clusters(self):
+        # all_boxes = self.csv2boxes()  # 获取所有box的宽和高
+        all_boxes = self.txt2boxes()
+        result = self.kmeans(all_boxes, k=self.cluster_number)
+        result = result[np.lexsort(result.T[0, None])]  # 对于聚类结果，按照第一维度进行从小到大排序
+        self.result2txt(result)
+        print("K anchors:\n {}".format(result))
+        print("Accuracy: {:.2f}%".format(
+            self.avg_iou(all_boxes, result) * 100))
+
+
+if __name__ == "__main__":
+    cluster_number = 6  # tiny-yolo--6, yolo--9
+    data_file = "labels/"
+    anchors_file = "anchors_6.txt"
+    kmeans = YOLO_Kmeans(cluster_number=cluster_number, data_file=data_file, anchors_file=anchors_file)
+    kmeans.txt2clusters()

bdd100k/anchors_6.txt

@@ -0,0 +1 @@
+3,7, 5,18, 7,10, 12,20, 26,38, 70,96

bdd100k/anchors_9.txt

@@ -0,0 +1 @@
+3,7, 5,18, 6,9, 10,32, 11,14, 17,21, 24,36, 45,59, 93,132

bdd100k/bdd100k.torrent

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23cadae5f079dcb12316ee7b6ba4542fb78ae6322ebdba90ae0fd6ef74a09db7
+size 18959009