From 4e5a00fb72335ab393b7a45903f7312ba33e773a Mon Sep 17 00:00:00 2001
From: Glenn Jocher <glenn.jocher@ultralytics.com>
Date: Mon, 15 Jul 2019 16:27:13 +0200
Subject: [PATCH] updates

---
 cfg/yolov3-spp-1cls.cfg |  2 +-
 test.py                 |  3 +++
 utils/utils.py          | 13 +++++++++----
 3 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/cfg/yolov3-spp-1cls.cfg b/cfg/yolov3-spp-1cls.cfg
index 3e633973..9e989e91 100644
--- a/cfg/yolov3-spp-1cls.cfg
+++ b/cfg/yolov3-spp-1cls.cfg
@@ -4,7 +4,7 @@
 # subdivisions=1
 # Training
 batch=64
-subdivisions=64
+subdivisions=32
 width=608
 height=608
 channels=3
diff --git a/test.py b/test.py
index e1563d26..051f5fc7 100644
--- a/test.py
+++ b/test.py
@@ -88,6 +88,9 @@ def test(
                     stats.append(([], torch.Tensor(), torch.Tensor(), tcls))
                 continue
 
+            # Clip boxes to image bounds
+            clip_coords(pred, shapes[si])
+
             # Append to text file
             # with open('test.txt', 'a') as file:
             #    [file.write('%11.5g' * 7 % tuple(x) + '\n') for x in pred]
diff --git a/utils/utils.py b/utils/utils.py
index 7fda5289..6f492fcf 100755
--- a/utils/utils.py
+++ b/utils/utils.py
@@ -125,16 +125,21 @@ def xywh2xyxy(x):
 
 
 def scale_coords(img1_shape, coords, img0_shape):
-    # Rescale coords1 (xyxy) from img1_shape to img0_shape
+    # Rescale coords (xyxy) from img1_shape to img0_shape
     gain = max(img1_shape) / max(img0_shape)  # gain  = old / new
     coords[:, [0, 2]] -= (img1_shape[1] - img0_shape[1] * gain) / 2  # x padding
     coords[:, [1, 3]] -= (img1_shape[0] - img0_shape[0] * gain) / 2  # y padding
     coords[:, :4] /= gain
-    coords[:, [0, 2]] = coords[:, [0, 2]].clamp(min=0, max=img0_shape[1])  # clip x
-    coords[:, [1, 3]] = coords[:, [1, 3]].clamp(min=0, max=img0_shape[0])  # clip y
+    clip_coords(coords, img0_shape)
     return coords
 
 
+def clip_coords(boxes, img_shape):
+    # Clip bounding xyxy bounding boxes to image shape (height, width)
+    boxes[:, [0, 2]] = boxes[:, [0, 2]].clamp(min=0, max=img_shape[1])  # clip x
+    boxes[:, [1, 3]] = boxes[:, [1, 3]].clamp(min=0, max=img_shape[0])  # clip y
+
+
 def ap_per_class(tp, conf, pred_cls, target_cls):
     """ Compute the average precision, given the recall and precision curves.
     Source: https://github.com/rafaelpadilla/Object-Detection-Metrics.
@@ -582,7 +587,7 @@ def kmeans_targets(path='./data/coco_64img.txt'):  # from utils.utils import *;
 # Plotting functions ---------------------------------------------------------------------------------------------------
 def plot_one_box(x, img, color=None, label=None, line_thickness=None):
     # Plots one bounding box on image img
-    tl = line_thickness or round(0.002 * (img.shape[0]+img.shape[1])/2) + 1  # line thickness
+    tl = line_thickness or round(0.002 * (img.shape[0] + img.shape[1]) / 2) + 1  # line thickness
     color = color or [random.randint(0, 255) for _ in range(3)]
     c1, c2 = (int(x[0]), int(x[1])), (int(x[2]), int(x[3]))
     cv2.rectangle(img, c1, c2, color, thickness=tl)