search

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.66k stars 7.96k forks source link

Detection of small objects #2310

Open malhosary opened 5 years ago

malhosary commented 5 years ago

@AlexeyAB Hi,

Thank you for the great effort and work. We are trying to detect multiple small objects (red marked rectangles in the image) (this is a sample image just to show you the size of objects that we want to detect) cam2 20190121_131948 1 _frame_1

Data Set:

Questions: 1- Which cfgfile you recommend to use for training? 2- What is the recommended width and height to use for training and detection? 3- How to calculate correct anchors for my data set? using: darknet detector calc_anchors data/obj.data -num_of_clusters 15 -final_width 16 -final_height 16 -width 832 -height 832 -show Or using: darknet detector calc_anchors data/obj.data -num_of_clusters 4 -width 832 -height 832 -show 4- What is the num_of_clusters , final_width and final_height and how to calculate it for my data set?

AlexeyAB commented 5 years ago

@malhosary Hi,

1/2. I would recommend you to train yolov3.cfg with width=608 height=608 or better width=832 height=832, if your GPU have enough GPU-RAM and it allows to achive to reach the speed you need.

  1. Use this command to calculate anchors, and attach the generated cloud of pointes to your message darknet detector calc_anchors data/obj.data -num_of_clusters 9 -width 832 -height 832 -show Write what anchors you get.

  2. -num_of_clusters 9 - number of anchors -width 832 -height 832 - input network size -final_width 16 -final_height 16 - deprecated and absent

Also read: https://github.com/AlexeyAB/darknet#how-to-improve-object-detection

recalculate anchors for your dataset for width and height from cfg-file: darknet.exe detector calc_anchors data/obj.data -num_of_clusters 9 -width 416 -height 416 then set the same 9 anchors in each of 3 [yolo]-layers in your cfg-file

(also you can change indexes of anchors masks= for each [yolo]-layer, so that 1st-[yolo]-layer has anchors larger than 60x60, 2nd larger than 30x30, 3rd remaining)

Use the latest version of this repository.

And train with flag -map darknet detector train data/obj.data yolo-obj.cfg darknet53.conv.74 -map Train at least 6000 iterations or more.

850 images for training and 200 images for validation

Also I recommend you to use 2x - 12x more training images if possible.

malhosary commented 5 years ago

@AlexeyAB Thanks for your quick reply. clusters

And this is the output for the command:

num_of_clusters = 9, width = 832, height = 832
 read labels from 686 images
 loaded          image: 686      box: 5618
 all loaded.

 calculating k-means++ ...

 avg IoU = 88.12 %

Saving anchors to the file: anchors.txt
anchors =   0,  0,   0,  0,  22, 43,  19, 73,  22, 67,  26, 65,  32, 69, 419,241, 570,391
AlexeyAB commented 5 years ago

Try to download the latest version of Darknet from this repository, and calc_anchors again. I fixed anchors calculation.

Also it looks like you have 100-200 large objects, that occupy ~50%x50% of image, do you want to detect such large objects?

malhosary commented 5 years ago

No, i want to detect only small products in the refrigerator

AlexeyAB commented 5 years ago

So you shouldn't use such big objects in your training dataset.

51906396-bc3e2280-23cc-11e9-80d7-94f5ee9e1a27

anchors = 0, 0, 0, 0, 22, 43, 19, 73, 22, 67, 26, 65, 32, 69, 419,241, 570,391

Also there shouldn't be 0 in the anchors. It was a bug.

malhosary commented 5 years ago

Hers are the new anchors using the new version:

new clusters

 num_of_clusters = 9, width = 832, height = 832
 read labels from 686 images
 loaded          image: 686      box: 5618
 all loaded.

 calculating k-means++ ...

 iterations = 3

 avg IoU = 91.67 %

Saving anchors to the file: anchors.txt
anchors =  20, 41,  22, 40,  23, 49,  21, 55,  20, 76,  29, 60,  26, 67,  32, 73, 496,318
dreambit commented 5 years ago

@AlexeyAB , you've said Also add max=200 to each [yolo]-layer. but in doc for training with a large number of objects in each image, add the parameter max=200 or higher value in the last [yolo]-layer

which one is correct?

AlexeyAB commented 5 years ago

@dreambit

You should to add max=200 in the last 3rd [yolo]-layer. But in this case it will be better to add max=200 and for 1st and 2nd yolo-layers.

AlexeyAB commented 5 years ago

@malhosary

You should remove such big objects from your training dataset, that occupys more than 25% of image, and then reclalculate anchors and run training.

51906396-bc3e2280-23cc-11e9-80d7-94f5ee9e1a27

dreambit commented 5 years ago

@AlexeyAB, i have to detect big objects and small objects(car) example: car1 000000000094 cars2 cars3

What is your recommendation for these objects? Is it enough to use default yolov3 config with custom generated anchors? or i have to use

for training for both small and large objects use modified models:

Full-model: 5 yolo layers: >https://raw.githubusercontent.com/AlexeyAB/darknet/master/cfg/yolov3_5l.cfg Tiny-model: 3 yolo layers: https://raw.githubusercontent.com/AlexeyAB/darknet/master/cfg/yolov3-tiny_3l.cfg Spatial-full-model: 3 yolo layers: >https://raw.githubusercontent.com/AlexeyAB/darknet/master/cfg/yolov3-spp.cfg

Thanks.

AlexeyAB commented 5 years ago

@dreambit It is better to use yolov3_5l.cfg in your case with default anchors, but it can be slow.

malhosary commented 5 years ago

@AlexeyAB Hi,

Thanks for your recommendations, i have here some good results aftertraining for 10 000 iterations with mAP: 94% and avg loss: 0.5

Green boxes are items that can be detected by my model, the issue is in Red boxes where the same items are in different locations but can't be detected (the same items with the same size but different location/position than locations in the training data set), what is your recommendations? 1_yolo

AlexeyAB commented 5 years ago

@malhosary

Yolo v3 can detect such objects with ~99.99% accuracy.

Just use more training images, about 2000 images for each class. And check that each object is labeled on your training images.

https://github.com/AlexeyAB/darknet#how-to-improve-object-detection

check that each object are mandatory labeled in your dataset - no one object in your data set should not be without label. In the most training issues - there are wrong labels in your dataset (got labels by using some conversion script, marked with a third-party tool, ...). Always check your dataset by using: https://github.com/AlexeyAB/Yolo_mark

guangwei commented 5 years ago

Hello, @AlexeyAB , I train very large picture (3644 X 2478) directly, and want to detect small objects (some are 4X4 pixels) from these large pictures. I tried to use yolov3-tiny.cfg and change width&height to 1920*1920, recalculate the anchors while I don't know if this anchor value is good, please check the attached picture, the inference result is not good, what's your recommendations to improve this ?

output of calculate anchors: num_of_clusters = 9, width = 1920, height = 1920 read labels from 201 images loaded image: 201 box: 1554 all loaded.

calculating k-means++ ...

avg IoU = 67.99 %

Saving anchors to the file: anchors.txt anchors = 12, 17, 14, 41, 34, 19, 25, 55, 50, 35, 18,111, 36, 79, 116, 27, 90, 69

anchors

AlexeyAB commented 5 years ago

@guangwei

Can you show anchors and point cloud if you calculate anchors for -width 3616 -height 2464 (values multiple of 32)?

guangwei commented 5 years ago

these are the annotations 0 0.244405021834 0.704148471616 0.00900655021834 0.0101892285298 0 0.453602620087 0.314956331878 0.0158296943231 0.0200145560408 0 0.547216157205 0.342066957787 0.0163755458515 0.0218340611354 0 0.544350436681 0.583333333333 0.0161026200873 0.0152838427948

I think the output of anchors is strange: ./darknet detector calc_anchors VOCdevkit/my.data -num_of_clusters 9 -width 3616 -height 2464 -show calc_anchors start

num_of_clusters = 9, width = 3616, height = 2464 read labels from 201 images loaded image: 201 box: 1554 all loaded.

calculating k-means++ ...

avg IoU = 68.13 %

Saving anchors to the file: anchors.txt anchors = 23, 22, 26, 51, 63, 24, 47, 70, 95, 44, 33,142, 67,101, 219, 35, 169, 88 anchors 1

AlexeyAB commented 5 years ago

@guangwei

So try to train yolov3-tiny.cfg with width = 3616 height = 2464 batch=64 subdivisions=64 in cfg-file.

and these anchors = 23, 22, 26, 51, 63, 24, 47, 70, 95, 44, 33,142, 67,101, 219, 35, 169, 88 in each of 3 yolo-layers.

wting861006 commented 5 years ago

@AlexeyAB Hi, about this image,can you give me some advice？This is a picture of sperm from an electron microscope. 3-14-3-20

AlexeyAB commented 5 years ago

@wting861006 Hi,

Show me output anchors and cloud.png darknet detector calc_anchors data/obj.data -num_of_clusters 9 -width 832 -height 832 -show

wting861006 commented 5 years ago

@wting861006 Hi,

Show me output anchors and cloud.png darknet detector calc_anchors data/obj.data -num_of_clusters 9 -width 832 -height 832 -show

image

AlexeyAB commented 5 years ago

@wting861006 So you should train yolov3.cfg or yolov3-tiny_3l.cfg with width=1024 height=1024 in cfg-file

and recalculated anchors darknet detector calc_anchors data/obj.data -num_of_clusters 9 -width 1024 -height 1024 -show

wting861006 commented 5 years ago

@wting861006 So you should train yolov3.cfg or yolov3-tiny_3l.cfg with width=1024 height=1024 in cfg-file

and recalculated anchors darknet detector calc_anchors data/obj.data -num_of_clusters 9 -width 1024 -height 1024 -show

I try yolov3-tiny_3l.cfg,yolov3-spp.cfg,yolov3.cfg,yolov3-tyny_5l.cfg, width=1280 height=1280.But the effect is not good.I upload my trained data in this place,could you please tell me what the problem might be?many thanks. trainingdata.zip

I use command:darknet detector train yolov3-tiny_3l.data yolov3-tiny_3l.cfg yolov3-tiny_3l_last.weights -gpus 0,1 -map -clear `Region 16 Avg IOU: 0.692852, Class: 0.999723, Obj: 0.653182, No Obj: 0.028169, .5R: 0.869565, .75R: 0.469565, count: 115 Region 16 Avg IOU: 0.694693, Class: 0.999744, Obj: 0.656049, No Obj: 0.028376, .5R: 0.890000, .75R: 0.450000, count: 100 Region 23 Avg IOU: 0.702654, Class: 0.999788, Obj: 0.421438, No Obj: 0.001485, .5R: 0.941176, .75R: 0.352941, count: 17 Region 23 Avg IOU: 0.771926, Class: 0.999771, Obj: 0.598884, No Obj: 0.008508, .5R: 0.954545, .75R: 0.645455, count: 110 Region 30 Avg IOU: 0.714068, Class: 0.999623, Obj: 0.553517, No Obj: 0.002131, .5R: 0.930435, .75R: 0.400000, count: 115 Region 16 Avg IOU: 0.673378, Class: 0.999779, Obj: 0.622265, No Obj: 0.014256, .5R: 0.865385, .75R: 0.423077, count: 52 Region 30 Avg IOU: 0.763364, Class: 0.999440, Obj: 0.613998, No Obj: 0.002058, .5R: 0.973451, .75R: 0.681416, count: 113 Region 23 Avg IOU: 0.762509, Class: 0.999860, Obj: 0.494922, No Obj: 0.004321, .5R: 0.981132, .75R: 0.603774, count: 53 Region 16 Avg IOU: 0.694208, Class: 0.999713, Obj: 0.685896, No Obj: 0.027605, .5R: 0.887640, .75R: 0.483146, count: 89 Region 23 Avg IOU: 0.769485, Class: 0.999680, Obj: 0.619020, No Obj: 0.009675, .5R: 0.954545, .75R: 0.689394, count: 132 Region 30 Avg IOU: 0.743331, Class: 0.999843, Obj: 0.627433, No Obj: 0.003065, .5R: 0.958824, .75R: 0.535294, count: 170 Region 30 Avg IOU: 0.768199, Class: 0.999487, Obj: 0.570825, No Obj: 0.001791, .5R: 0.954545, .75R: 0.727273, count: 110 Syncing... Done!

(next mAP calculation at 2452 iterations) Last accuracy mAP@0.5 = 19.26 % 2456: 64.943108, 68.233200 avg loss, 0.000020 rate, 1.354000 seconds, 9824 images

calculation mAP (mean average precision)... 3 detections_count = 3154, unique_truth_count = 1204 class_id = 0, name = sperm, ap = 46.22% (TP = 573, FP = 391) class_id = 1, name = round cell, ap = 0.00% (TP = 0, FP = 0) class_id = 2, name = red cell, ap = 0.00% (TP = 0, FP = 0)

for thresh = 0.25, precision = 0.59, recall = 0.48, F1-score = 0.53 for thresh = 0.25, TP = 573, FP = 391, FN = 631, average IoU = 37.42 %

IoU threshold = 50 %, used Area-Under-Curve for each unique Recall mean average precision (mAP@0.50) = 0.154082, or 15.41 % Total Detection Time: 1.000000 Seconds

Set -points flag: -points 101 for MS COCO -points 11 for PascalVOC 2007 (uncomment difficult in voc.data) -points 0 (AUC) for ImageNet, PascalVOC 2010-2012, your custom dataset

mean_average_precision (mAP@0.5) = 0.154082 Loaded: 0.000000 seconds`

AlexeyAB commented 5 years ago

@wting861006 Hi,

calculation mAP (mean average precision)... 3 detections_count = 3154, unique_truth_count = 1204 class_id = 0, name = sperm, ap = 46.22% (TP = 573, FP = 391) class_id = 1, name = round cell, ap = 0.00% (TP = 0, FP = 0) class_id = 2, name = red cell, ap = 0.00% (TP = 0, FP = 0)

for thresh = 0.25, precision = 0.59, recall = 0.48, F1-score = 0.53 for thresh = 0.25, TP = 573, FP = 391, FN = 631, average IoU = 37.42 %

IoU threshold = 50 %, used Area-Under-Curve for each unique Recall mean average precision (mAP@0.50) = 0.154082, or 15.41 %

Your labels look like good, but check that you labeled all of (sperm, round cell, red cell) objects on the image without any exception: image

Train at least 6000 iterations by using this cfg-file without any changes (just if out of memory occurs then set random=0 in the last yolo-layer): yolov3-tiny_3l.cfg.txt

Use: https://github.com/AlexeyAB/darknet#how-to-train-tiny-yolo-to-detect-your-custom-objects darknet detector train yolov3-tiny_3l.data yolov3-tiny_3l.cfg yolov3-tiny.conv.15 -map

After 2000 iterations you can run multi-gpu training: darknet detector train yolov3-tiny_3l.data yolov3-tiny_3l.cfg backup/yolov3-tiny_3l_2000.weights -gpus 0,1 -map

wting861006 commented 5 years ago

@wting861006 Hi,

calculation mAP (mean average precision)... 3 detections_count = 3154, unique_truth_count = 1204 class_id = 0, name = sperm, ap = 46.22% (TP = 573, FP = 391) class_id = 1, name = round cell, ap = 0.00% (TP = 0, FP = 0) class_id = 2, name = red cell, ap = 0.00% (TP = 0, FP = 0) for thresh = 0.25, precision = 0.59, recall = 0.48, F1-score = 0.53 for thresh = 0.25, TP = 573, FP = 391, FN = 631, average IoU = 37.42 % IoU threshold = 50 %, used Area-Under-Curve for each unique Recall mean average precision (mAP@0.50) = 0.154082, or 15.41 %

  • Actually you get not bad result mAP = 46.22% for 2500 iterations, just train more (you see mAP = 15.41% = 46.22/3 just because you don't have any images for class_id 1 & 2)
  • As I see you have only 3 validaton images, it is very low number.
  • Put labels in the same folder where are images
  • How many training images do you have?
  • For production - to use in real cases - you should have 200 - 2000 training images for each class: https://github.com/AlexeyAB/darknet#how-to-improve-object-detection
  • Don't set so low learning rate 0.00001

Your labels look like good, but check that you labeled all of (sperm, round cell, red cell) objects on the image without any exception: image

Train at least 6000 iterations by using this cfg-file without any changes (just if out of memory occurs then set random=0 in the last yolo-layer): yolov3-tiny_3l.cfg.txt

Use: https://github.com/AlexeyAB/darknet#how-to-train-tiny-yolo-to-detect-your-custom-objects darknet detector train yolov3-tiny_3l.data yolov3-tiny_3l.cfg yolov3-tiny.conv.15 -map

After 2000 iterations you can run multi-gpu training: darknet detector train yolov3-tiny_3l.data yolov3-tiny_3l.cfg backup/yolov3-tiny_3l_2000.weights -gpus 0,1 -map

@AlexeyAB Hi,I trained according to your suggestion. Except for the high loss, it seemed to be fine.

(next mAP calculation at 8040 iterations) Last accuracy mAP@0.5 = 23.18 % 7999: 28.282316, 28.417795 avg loss, 0.000200 rate, 32.724000 seconds, 1023872 images Loaded: 0.001000 seconds Region 16 Avg IOU: 0.728459, Class: 0.999942, Obj: 0.871590, No Obj: 0.027605, .5R: 0.916667, .75R: 0.500000, count: 24 Region 16 Avg IOU: 0.789409, Class: 0.999979, Obj: 0.855757, No Obj: 0.037393, .5R: 0.970588, .75R: 0.705882, count: 34 Region 23 Avg IOU: 0.728356, Class: 0.999971, Obj: 0.852199, No Obj: 0.010514, .5R: 0.925373, .75R: 0.582090, count: 134 Region 23 Avg IOU: 0.831161, Class: 0.999964, Obj: 0.903209, No Obj: 0.014844, .5R: 0.985714, .75R: 0.835714, count: 140 Region 30 Avg IOU: 0.727796, Class: 0.999994, Obj: 0.706936, No Obj: 0.001159, .5R: 0.905660, .75R: 0.566038, count: 53 Region 16 Avg IOU: 0.791163, Class: 0.999981, Obj: 0.865418, No Obj: 0.044673, .5R: 0.911111, .75R: 0.755556, count: 45 Region 30 Avg IOU: 0.821844, Class: 0.999983, Obj: 0.858697, No Obj: 0.004129, .5R: 0.959016, .75R: 0.827869, count: 122 Region 23 Avg IOU: 0.815609, Class: 0.999944, Obj: 0.864114, No Obj: 0.016482, .5R: 0.978571, .75R: 0.828571, count: 140 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000105, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.757245, Class: 0.999986, Obj: 0.819335, No Obj: 0.002877, .5R: 0.928571, .75R: 0.678571, count: 28 Region 30 Avg IOU: 0.836744, Class: 0.999990, Obj: 0.870285, No Obj: 0.004231, .5R: 0.982759, .75R: 0.870690, count: 116 Region 16 Avg IOU: 0.758645, Class: 0.999965, Obj: 0.888311, No Obj: 0.053629, .5R: 0.921569, .75R: 0.686275, count: 51 Region 23 Avg IOU: 0.822879, Class: 0.999961, Obj: 0.917018, No Obj: 0.015650, .5R: 0.992857, .75R: 0.842857, count: 140 Region 30 Avg IOU: 0.746448, Class: 0.999994, Obj: 0.850215, No Obj: 0.007791, .5R: 0.920000, .75R: 0.608571, count: 350 Region 16 Avg IOU: 0.817286, Class: 0.999964, Obj: 0.893398, No Obj: 0.020972, .5R: 1.000000, .75R: 0.777778, count: 18 Region 23 Avg IOU: 0.804021, Class: 0.999966, Obj: 0.889353, No Obj: 0.017437, .5R: 0.961538, .75R: 0.774725, count: 182 Region 30 Avg IOU: 0.756729, Class: 0.999978, Obj: 0.756026, No Obj: 0.002091, .5R: 0.857143, .75R: 0.698413, count: 63 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000008, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.790081, Class: 0.999975, Obj: 0.821583, No Obj: 0.005539, .5R: 0.963636, .75R: 0.745455, count: 55 Region 30 Avg IOU: 0.823043, Class: 0.999965, Obj: 0.858511, No Obj: 0.004140, .5R: 0.961165, .75R: 0.844660, count: 103 Region 16 Avg IOU: 0.708436, Class: 0.999981, Obj: 0.865744, No Obj: 0.054155, .5R: 0.843137, .75R: 0.568627, count: 51 Region 23 Avg IOU: 0.817897, Class: 0.999963, Obj: 0.882496, No Obj: 0.014603, .5R: 0.971429, .75R: 0.821429, count: 140 Region 30 Avg IOU: 0.779259, Class: 0.999994, Obj: 0.855687, No Obj: 0.007194, .5R: 0.950658, .75R: 0.703947, count: 304 Region 16 Avg IOU: 0.800896, Class: 0.999968, Obj: 0.882048, No Obj: 0.025618, .5R: 1.000000, .75R: 0.772727, count: 22 Region 30 Avg IOU: 0.825269, Class: 0.999958, Obj: 0.858482, No Obj: 0.002874, .5R: 0.926471, .75R: 0.852941, count: 68 Region 23 Avg IOU: 0.817418, Class: 0.999966, Obj: 0.867868, No Obj: 0.015459, .5R: 0.987097, .75R: 0.819355, count: 155 Region 16 Avg IOU: 0.736228, Class: 0.999954, Obj: 0.822897, No Obj: 0.067031, .5R: 0.954545, .75R: 0.500000, count: 66 Region 23 Avg IOU: 0.715319, Class: 0.999950, Obj: 0.736167, No Obj: 0.008709, .5R: 0.921569, .75R: 0.509804, count: 102 Region 30 Avg IOU: 0.673622, Class: 0.999986, Obj: 0.586671, No Obj: 0.000254, .5R: 0.818182, .75R: 0.545455, count: 11 Region 16 Avg IOU: 0.777791, Class: 0.999965, Obj: 0.925473, No Obj: 0.036930, .5R: 0.965517, .75R: 0.724138, count: 29 Region 30 Avg IOU: 0.843717, Class: 0.999981, Obj: 0.910572, No Obj: 0.005666, .5R: 0.988506, .75R: 0.896552, count: 174 Region 23 Avg IOU: 0.757773, Class: 0.999961, Obj: 0.871485, No Obj: 0.012311, .5R: 0.948905, .75R: 0.649635, count: 137 Region 16 Avg IOU: 0.789821, Class: 0.999988, Obj: 0.832196, No Obj: 0.021670, .5R: 1.000000, .75R: 0.700000, count: 20 Region 23 Avg IOU: 0.818793, Class: 0.999945, Obj: 0.851732, No Obj: 0.016416, .5R: 0.974684, .75R: 0.803797, count: 158 Region 30 Avg IOU: 0.675039, Class: 0.999972, Obj: 0.710975, No Obj: 0.000938, .5R: 0.678571, .75R: 0.535714, count: 28 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.001054, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.758965, Class: 0.999989, Obj: 0.869013, No Obj: 0.009182, .5R: 0.953271, .75R: 0.579439, count: 107 Region 30 Avg IOU: 0.841728, Class: 0.999973, Obj: 0.907757, No Obj: 0.008206, .5R: 0.983806, .75R: 0.890688, count: 247 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000098, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 30 Avg IOU: 0.776824, Class: 0.999994, Obj: 0.801551, No Obj: 0.005489, .5R: 0.934579, .75R: 0.719626, count: 214 Region 16 Avg IOU: 0.784695, Class: 0.999976, Obj: 0.921864, No Obj: 0.054677, .5R: 0.979592, .75R: 0.693878, count: 49 Region 23 Avg IOU: 0.837545, Class: 0.999980, Obj: 0.940718, No Obj: 0.005687, .5R: 1.000000, .75R: 0.916667, count: 48 Region 23 Avg IOU: 0.784718, Class: 0.999907, Obj: 0.797513, No Obj: 0.015029, .5R: 0.940476, .75R: 0.750000, count: 168 Region 30 Avg IOU: 0.819371, Class: 0.999931, Obj: 0.830734, No Obj: 0.002151, .5R: 0.944444, .75R: 0.814815, count: 54 Region 16 Avg IOU: 0.788925, Class: 0.999984, Obj: 0.945039, No Obj: 0.018906, .5R: 0.933333, .75R: 0.666667, count: 15 Region 30 Avg IOU: 0.822104, Class: 0.999988, Obj: 0.909026, No Obj: 0.010283, .5R: 0.967302, .75R: 0.839237, count: 367 Region 23 Avg IOU: 0.827880, Class: 0.999988, Obj: 0.912221, No Obj: 0.013392, .5R: 0.990909, .75R: 0.845455, count: 110 Region 16 Avg IOU: 0.796615, Class: 0.999987, Obj: 0.961586, No Obj: 0.017919, .5R: 1.000000, .75R: 0.818182, count: 11 Region 23 Avg IOU: 0.796908, Class: 0.999990, Obj: 0.926985, No Obj: 0.012506, .5R: 0.929688, .75R: 0.796875, count: 128 Region 30 Avg IOU: 0.840303, Class: 0.999994, Obj: 0.927013, No Obj: 0.006714, .5R: 0.995169, .75R: 0.859903, count: 207 Region 30 Avg IOU: 0.824875, Class: 0.999990, Obj: 0.898489, No Obj: 0.004171, .5R: 0.985075, .75R: 0.865672, count: 134 Region 16 Avg IOU: 0.690382, Class: 0.999947, Obj: 0.848882, No Obj: 0.092834, .5R: 0.855556, .75R: 0.422222, count: 90 Region 16 Avg IOU: 0.826808, Class: 0.999971, Obj: 0.945638, No Obj: 0.009363, .5R: 1.000000, .75R: 0.875000, count: 8 Region 23 Avg IOU: 0.749426, Class: 0.999720, Obj: 0.699263, No Obj: 0.006849, .5R: 0.955556, .75R: 0.644444, count: 90 Region 23 Avg IOU: 0.844091, Class: 0.999984, Obj: 0.941422, No Obj: 0.012538, .5R: 1.000000, .75R: 0.858586, count: 99 Region 30 Avg IOU: 0.632242, Class: 0.999931, Obj: 0.518282, No Obj: 0.000337, .5R: 0.777778, .75R: 0.222222, count: 9 Region 16 Avg IOU: 0.754437, Class: 0.999987, Obj: 0.881478, No Obj: 0.025438, .5R: 0.920000, .75R: 0.720000, count: 25 Region 23 Avg IOU: 0.838224, Class: 0.999958, Obj: 0.910537, No Obj: 0.018389, .5R: 0.970930, .75R: 0.848837, count: 172 Region 30 Avg IOU: 0.832407, Class: 0.999986, Obj: 0.928840, No Obj: 0.008180, .5R: 0.985240, .75R: 0.841328, count: 271 Region 16 Avg IOU: 0.800871, Class: 0.999975, Obj: 0.927815, No Obj: 0.010643, .5R: 1.000000, .75R: 0.800000, count: 10 Region 23 Avg IOU: 0.787905, Class: 0.999957, Obj: 0.903203, No Obj: 0.012896, .5R: 0.924528, .75R: 0.773585, count: 106 Region 30 Avg IOU: 0.831920, Class: 0.999983, Obj: 0.908863, No Obj: 0.005794, .5R: 0.961290, .75R: 0.864516, count: 155 Region 16 Avg IOU: 0.720263, Class: 0.999958, Obj: 0.844352, No Obj: 0.049831, .5R: 0.886364, .75R: 0.545455, count: 44 Region 23 Avg IOU: 0.741264, Class: 0.999968, Obj: 0.757226, No Obj: 0.009297, .5R: 0.918033, .75R: 0.590164, count: 122 Region 30 Avg IOU: 0.833528, Class: 0.999974, Obj: 0.901680, No Obj: 0.006262, .5R: 0.963351, .75R: 0.853403, count: 191 Region 16 Avg IOU: 0.722023, Class: 0.999967, Obj: 0.846855, No Obj: 0.052589, .5R: 0.921569, .75R: 0.529412, count: 51 Region 30 Avg IOU: 0.731805, Class: 0.999984, Obj: 0.676473, No Obj: 0.000944, .5R: 0.923077, .75R: 0.589744, count: 39 Region 23 Avg IOU: 0.747586, Class: 0.999965, Obj: 0.808995, No Obj: 0.016198, .5R: 0.934426, .75R: 0.606557, count: 183 Region 16 Avg IOU: 0.779799, Class: 0.999982, Obj: 0.854611, No Obj: 0.016208, .5R: 1.000000, .75R: 0.529412, count: 17 Region 23 Avg IOU: 0.785591, Class: 0.999984, Obj: 0.886845, No Obj: 0.015702, .5R: 0.968354, .75R: 0.734177, count: 158 Region 30 Avg IOU: 0.749853, Class: 0.999991, Obj: 0.741158, No Obj: 0.000832, .5R: 0.913043, .75R: 0.608696, count: 23 Region 16 Avg IOU: 0.764985, Class: 0.999960, Obj: 0.833427, No Obj: 0.038541, .5R: 0.906250, .75R: 0.718750, count: 32 Region 23 Avg IOU: 0.743274, Class: 0.999985, Obj: 0.821249, No Obj: 0.009999, .5R: 0.910569, .75R: 0.609756, count: 123 Region 30 Avg IOU: 0.804083, Class: 0.999985, Obj: 0.893455, No Obj: 0.003931, .5R: 0.966387, .75R: 0.764706, count: 119 Region 16 Avg IOU: 0.763691, Class: 0.999979, Obj: 0.893420, No Obj: 0.021691, .5R: 1.000000, .75R: 0.600000, count: 20 Region 23 Avg IOU: 0.834073, Class: 0.999990, Obj: 0.929929, No Obj: 0.012755, .5R: 0.991525, .75R: 0.872881, count: 118 Region 30 Avg IOU: 0.755934, Class: 0.999963, Obj: 0.655496, No Obj: 0.000669, .5R: 0.888889, .75R: 0.703704, count: 27 Region 16 Avg IOU: 0.878477, Class: 0.999975, Obj: 0.957937, No Obj: 0.011833, .5R: 1.000000, .75R: 1.000000, count: 6 Region 23 Avg IOU: 0.831948, Class: 0.999934, Obj: 0.921172, No Obj: 0.011971, .5R: 0.980769, .75R: 0.865385, count: 104 Region 30 Avg IOU: 0.839456, Class: 0.999987, Obj: 0.895580, No Obj: 0.003585, .5R: 0.967742, .75R: 0.881720, count: 93 Region 16 Avg IOU: 0.798397, Class: 0.999991, Obj: 0.967870, No Obj: 0.009056, .5R: 1.000000, .75R: 0.600000, count: 5 Region 23 Avg IOU: 0.819347, Class: 0.999958, Obj: 0.889570, No Obj: 0.013659, .5R: 0.984733, .75R: 0.854962, count: 131 Region 30 Avg IOU: 0.835097, Class: 0.999989, Obj: 0.922987, No Obj: 0.008175, .5R: 0.992509, .75R: 0.868914, count: 267 Region 16 Avg IOU: 0.761256, Class: 0.999963, Obj: 0.870063, No Obj: 0.035201, .5R: 1.000000, .75R: 0.625000, count: 32 Region 23 Avg IOU: 0.757937, Class: 0.999963, Obj: 0.838941, No Obj: 0.012911, .5R: 0.904459, .75R: 0.656051, count: 157 Region 30 Avg IOU: 0.831408, Class: 0.999986, Obj: 0.912502, No Obj: 0.007425, .5R: 0.965517, .75R: 0.862069, count: 232 Region 16 Avg IOU: 0.872516, Class: 0.999996, Obj: 0.968698, No Obj: 0.001569, .5R: 1.000000, .75R: 1.000000, count: 1 Region 23 Avg IOU: 0.831201, Class: 0.999980, Obj: 0.847159, No Obj: 0.005075, .5R: 1.000000, .75R: 0.823529, count: 34 Region 30 Avg IOU: 0.772045, Class: 0.999981, Obj: 0.742372, No Obj: 0.000761, .5R: 0.928571, .75R: 0.607143, count: 28 Region 16 Avg IOU: 0.748502, Class: 0.999989, Obj: 0.782190, No Obj: 0.004269, .5R: 1.000000, .75R: 0.666667, count: 3 Region 23 Avg IOU: 0.757831, Class: 0.999986, Obj: 0.854116, No Obj: 0.012950, .5R: 0.912162, .75R: 0.662162, count: 148 Region 30 Avg IOU: 0.793263, Class: 0.999996, Obj: 0.892331, No Obj: 0.007779, .5R: 0.964630, .75R: 0.758842, count: 311 Region 16 Avg IOU: 0.768529, Class: 0.999971, Obj: 0.919163, No Obj: 0.069179, .5R: 0.933333, .75R: 0.716667, count: 60 Region 23 Avg IOU: 0.794949, Class: 0.999985, Obj: 0.867421, No Obj: 0.013788, .5R: 0.967320, .75R: 0.745098, count: 153 Region 30 Avg IOU: 0.808955, Class: 0.999987, Obj: 0.879069, No Obj: 0.005687, .5R: 0.957447, .75R: 0.803191, count: 188 Region 16 Avg IOU: 0.857313, Class: 0.999990, Obj: 0.393047, No Obj: 0.000647, .5R: 1.000000, .75R: 1.000000, count: 1 Region 23 Avg IOU: 0.817555, Class: 0.999979, Obj: 0.870660, No Obj: 0.003504, .5R: 1.000000, .75R: 0.782609, count: 23 Region 30 Avg IOU: 0.785536, Class: 0.999971, Obj: 0.706055, No Obj: 0.001377, .5R: 0.976190, .75R: 0.690476, count: 42 Region 16 Avg IOU: 0.813957, Class: 0.999990, Obj: 0.921338, No Obj: 0.017380, .5R: 1.000000, .75R: 0.785714, count: 14 Region 23 Avg IOU: 0.802159, Class: 0.999948, Obj: 0.865003, No Obj: 0.015187, .5R: 0.968944, .75R: 0.782609, count: 161 Region 30 Avg IOU: 0.789212, Class: 0.999994, Obj: 0.910847, No Obj: 0.009042, .5R: 0.953039, .75R: 0.718232, count: 362 Region 16 Avg IOU: 0.765696, Class: 0.999980, Obj: 0.961934, No Obj: 0.006056, .5R: 1.000000, .75R: 0.600000, count: 5 Region 23 Avg IOU: 0.809739, Class: 0.999994, Obj: 0.896761, No Obj: 0.006397, .5R: 0.984375, .75R: 0.796875, count: 64 Region 30 Avg IOU: 0.828939, Class: 0.999979, Obj: 0.863415, No Obj: 0.006236, .5R: 0.981221, .75R: 0.845070, count: 213 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.001079, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.778810, Class: 0.999992, Obj: 0.775562, No Obj: 0.002977, .5R: 1.000000, .75R: 0.695652, count: 23 Region 30 Avg IOU: 0.804343, Class: 0.999993, Obj: 0.884081, No Obj: 0.006174, .5R: 0.974359, .75R: 0.786325, count: 234 Region 16 Avg IOU: 0.796003, Class: 0.999980, Obj: 0.883842, No Obj: 0.025013, .5R: 0.958333, .75R: 0.708333, count: 24 Region 23 Avg IOU: 0.822314, Class: 0.999964, Obj: 0.904165, No Obj: 0.013927, .5R: 0.968000, .75R: 0.808000, count: 125 Region 30 Avg IOU: 0.776246, Class: 0.999993, Obj: 0.876647, No Obj: 0.008473, .5R: 0.964187, .75R: 0.705234, count: 363 Region 16 Avg IOU: 0.853101, Class: 0.999896, Obj: 0.840699, No Obj: 0.003954, .5R: 1.000000, .75R: 1.000000, count: 2 Region 23 Avg IOU: 0.778022, Class: 0.999961, Obj: 0.860394, No Obj: 0.012448, .5R: 0.962687, .75R: 0.716418, count: 134 Region 30 Avg IOU: 0.838394, Class: 0.999974, Obj: 0.895975, No Obj: 0.006170, .5R: 0.985294, .75R: 0.901961, count: 204 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000184, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.833173, Class: 0.999980, Obj: 0.913155, No Obj: 0.006219, .5R: 1.000000, .75R: 0.844828, count: 58 Region 30 Avg IOU: 0.815699, Class: 0.999993, Obj: 0.890091, No Obj: 0.006385, .5R: 0.967593, .75R: 0.810185, count: 216 Region 16 Avg IOU: 0.764944, Class: 0.999989, Obj: 0.921200, No Obj: 0.038511, .5R: 0.935484, .75R: 0.709677, count: 31 Region 23 Avg IOU: 0.792637, Class: 0.999948, Obj: 0.857670, No Obj: 0.014153, .5R: 0.969697, .75R: 0.757576, count: 132 Region 30 Avg IOU: 0.819246, Class: 0.999993, Obj: 0.920825, No Obj: 0.009363, .5R: 0.956923, .75R: 0.833846, count: 325 Region 16 Avg IOU: 0.803521, Class: 0.999977, Obj: 0.904919, No Obj: 0.053224, .5R: 1.000000, .75R: 0.745098, count: 51 Region 23 Avg IOU: 0.810103, Class: 0.999979, Obj: 0.890915, No Obj: 0.012620, .5R: 0.964286, .75R: 0.830357, count: 112 Region 30 Avg IOU: 0.830770, Class: 0.999968, Obj: 0.868687, No Obj: 0.003308, .5R: 0.979167, .75R: 0.885417, count: 96 Region 16 Avg IOU: 0.740964, Class: 0.999967, Obj: 0.854958, No Obj: 0.057035, .5R: 0.925926, .75R: 0.518519, count: 54 Region 23 Avg IOU: 0.755742, Class: 0.999931, Obj: 0.834745, No Obj: 0.010322, .5R: 0.938053, .75R: 0.637168, count: 113 Region 30 Avg IOU: 0.827658, Class: 0.999938, Obj: 0.875942, No Obj: 0.002322, .5R: 0.952381, .75R: 0.920635, count: 63 Region 16 Avg IOU: 0.757504, Class: 0.999974, Obj: 0.869383, No Obj: 0.055546, .5R: 0.924528, .75R: 0.603774, count: 53 Region 23 Avg IOU: 0.790637, Class: 0.999908, Obj: 0.842142, No Obj: 0.013648, .5R: 0.972028, .75R: 0.734266, count: 143 Region 30 Avg IOU: 0.742443, Class: 0.999960, Obj: 0.675435, No Obj: 0.001483, .5R: 0.919355, .75R: 0.661290, count: 62 Region 16 Avg IOU: 0.775539, Class: 0.999967, Obj: 0.871145, No Obj: 0.040769, .5R: 0.945946, .75R: 0.675676, count: 37 Region 23 Avg IOU: 0.779890, Class: 0.999971, Obj: 0.865131, No Obj: 0.015537, .5R: 0.923077, .75R: 0.737179, count: 156 Region 30 Avg IOU: 0.837252, Class: 0.999950, Obj: 0.820496, No Obj: 0.002040, .5R: 0.972973, .75R: 0.905405, count: 74 Region 16 Avg IOU: 0.771877, Class: 0.999973, Obj: 0.890585, No Obj: 0.021097, .5R: 0.944444, .75R: 0.611111, count: 18 Region 23 Avg IOU: 0.692963, Class: 0.999978, Obj: 0.786946, No Obj: 0.012610, .5R: 0.847561, .75R: 0.487805, count: 164 Region 30 Avg IOU: 0.793773, Class: 0.999967, Obj: 0.803041, No Obj: 0.002347, .5R: 0.958333, .75R: 0.708333, count: 72 Region 16 Avg IOU: 0.730396, Class: 0.999964, Obj: 0.947181, No Obj: 0.014562, .5R: 1.000000, .75R: 0.600000, count: 10 Region 23 Avg IOU: 0.729929, Class: 0.999987, Obj: 0.821545, No Obj: 0.015871, .5R: 0.882927, .75R: 0.570732, count: 205 Region 30 Avg IOU: 0.756141, Class: 0.999974, Obj: 0.615413, No Obj: 0.000946, .5R: 0.947368, .75R: 0.631579, count: 38 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000578, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.807425, Class: 0.999967, Obj: 0.681013, No Obj: 0.002491, .5R: 1.000000, .75R: 0.714286, count: 21 Region 30 Avg IOU: 0.759417, Class: 0.999969, Obj: 0.781542, No Obj: 0.001754, .5R: 0.918033, .75R: 0.672131, count: 61 Region 16 Avg IOU: 0.692464, Class: 0.999970, Obj: 0.742669, No Obj: 0.010214, .5R: 0.800000, .75R: 0.500000, count: 10 Region 23 Avg IOU: 0.786766, Class: 0.999970, Obj: 0.822876, No Obj: 0.014264, .5R: 0.958904, .75R: 0.732877, count: 146 Region 30 Avg IOU: 0.763948, Class: 0.999993, Obj: 0.827462, No Obj: 0.008469, .5R: 0.944598, .75R: 0.656510, count: 361 Region 16 Avg IOU: 0.799430, Class: 0.999984, Obj: 0.807299, No Obj: 0.009658, .5R: 1.000000, .75R: 0.800000, count: 10 Region 30 Avg IOU: 0.810641, Class: 0.999980, Obj: 0.873936, No Obj: 0.006066, .5R: 0.960784, .75R: 0.799020, count: 204 Region 23 Avg IOU: 0.792055, Class: 0.999969, Obj: 0.888835, No Obj: 0.012960, .5R: 0.968750, .75R: 0.726563, count: 128 Region 16 Avg IOU: 0.836822, Class: 0.999986, Obj: 0.880808, No Obj: 0.025861, .5R: 0.954545, .75R: 0.909091, count: 22 Region 23 Avg IOU: 0.821071, Class: 0.999991, Obj: 0.912370, No Obj: 0.014157, .5R: 0.975806, .75R: 0.798387, count: 124 Region 30 Avg IOU: 0.822604, Class: 0.999988, Obj: 0.861707, No Obj: 0.007797, .5R: 0.967857, .75R: 0.835714, count: 280 Region 30 Avg IOU: 0.828177, Class: 0.999988, Obj: 0.899844, No Obj: 0.005043, .5R: 0.947712, .75R: 0.895425, count: 153 Region 16 Avg IOU: 0.845334, Class: 0.999987, Obj: 0.961435, No Obj: 0.025912, .5R: 1.000000, .75R: 0.850000, count: 20 Region 16 Avg IOU: 0.748366, Class: 0.999984, Obj: 0.899828, No Obj: 0.041463, .5R: 0.916667, .75R: 0.611111, count: 36 Region 23 Avg IOU: 0.821120, Class: 0.999963, Obj: 0.892257, No Obj: 0.015993, .5R: 0.978417, .75R: 0.856115, count: 139 Region 23 Avg IOU: 0.842402, Class: 0.999962, Obj: 0.892871, No Obj: 0.015367, .5R: 0.985294, .75R: 0.904412, count: 136 Region 30 Avg IOU: 0.846450, Class: 0.999969, Obj: 0.886495, No Obj: 0.004609, .5R: 0.992188, .75R: 0.906250, count: 128 Region 16 Avg IOU: 0.672694, Class: 0.999962, Obj: 0.943238, No Obj: 0.006091, .5R: 1.000000, .75R: 0.250000, count: 4 Region 30 Avg IOU: 0.847050, Class: 0.999987, Obj: 0.899536, No Obj: 0.006868, .5R: 0.973404, .75R: 0.898936, count: 188 Region 16 Avg IOU: 0.688844, Class: 0.999965, Obj: 0.664333, No Obj: 0.013326, .5R: 0.928571, .75R: 0.428571, count: 14 Region 23 Avg IOU: 0.718276, Class: 0.999991, Obj: 0.817508, No Obj: 0.014432, .5R: 0.883249, .75R: 0.563452, count: 197 Region 23 Avg IOU: 0.725209, Class: 0.999950, Obj: 0.780371, No Obj: 0.010743, .5R: 0.908497, .75R: 0.542484, count: 153 Region 30 Avg IOU: 0.785260, Class: 0.999994, Obj: 0.808276, No Obj: 0.002375, .5R: 0.941176, .75R: 0.682353, count: 85 Region 16 Avg IOU: 0.777749, Class: 0.999969, Obj: 0.910252, No Obj: 0.066879, .5R: 0.950820, .75R: 0.672131, count: 61 Region 30 Avg IOU: 0.738516, Class: 0.999985, Obj: 0.681831, No Obj: 0.002636, .5R: 0.918033, .75R: 0.598361, count: 122 Region 23 Avg IOU: 0.797863, Class: 0.999964, Obj: 0.859742, No Obj: 0.012267, .5R: 0.945736, .75R: 0.790698, count: 129 Region 16 Avg IOU: 0.928579, Class: 0.999975, Obj: 0.997926, No Obj: 0.002298, .5R: 1.000000, .75R: 1.000000, count: 1 Region 23 Avg IOU: 0.803332, Class: 0.999971, Obj: 0.848549, No Obj: 0.006755, .5R: 0.967213, .75R: 0.803279, count: 61 Region 30 Avg IOU: 0.792241, Class: 0.999955, Obj: 0.760447, No Obj: 0.001448, .5R: 0.951219, .75R: 0.780488, count: 41 Region 16 Avg IOU: 0.741723, Class: 0.999963, Obj: 0.822805, No Obj: 0.069265, .5R: 0.897059, .75R: 0.588235, count: 68 Region 23 Avg IOU: 0.788396, Class: 0.999942, Obj: 0.816687, No Obj: 0.012999, .5R: 0.977273, .75R: 0.742424, count: 132 Region 30 Avg IOU: 0.806839, Class: 0.999992, Obj: 0.921134, No Obj: 0.007306, .5R: 0.960714, .75R: 0.785714, count: 280 Region 16 Avg IOU: 0.772473, Class: 0.999983, Obj: 0.865483, No Obj: 0.035309, .5R: 0.937500, .75R: 0.687500, count: 32 Region 23 Avg IOU: 0.760208, Class: 0.999947, Obj: 0.798656, No Obj: 0.010803, .5R: 0.931624, .75R: 0.623932, count: 117 Region 30 Avg IOU: 0.791559, Class: 0.999984, Obj: 0.808366, No Obj: 0.001942, .5R: 0.943396, .75R: 0.792453, count: 53 Region 16 Avg IOU: 0.824709, Class: 0.999984, Obj: 0.900123, No Obj: 0.031624, .5R: 1.000000, .75R: 0.916667, count: 24 Region 23 Avg IOU: 0.829783, Class: 0.999974, Obj: 0.928299, No Obj: 0.014739, .5R: 0.964286, .75R: 0.892857, count: 140 Region 30 Avg IOU: 0.813559, Class: 0.999965, Obj: 0.757607, No Obj: 0.001566, .5R: 0.976744, .75R: 0.813953, count: 43 Region 16 Avg IOU: 0.782935, Class: 0.999967, Obj: 0.827323, No Obj: 0.054261, .5R: 0.960784, .75R: 0.666667, count: 51 Region 23 Avg IOU: 0.783351, Class: 0.999974, Obj: 0.851599, No Obj: 0.012230, .5R: 0.929688, .75R: 0.750000, count: 128 Region 30 Avg IOU: 0.832326, Class: 0.999977, Obj: 0.877686, No Obj: 0.005439, .5R: 0.982036, .75R: 0.850299, count: 167 Region 30 Avg IOU: 0.818918, Class: 0.999992, Obj: 0.849400, No Obj: 0.001754, .5R: 0.981132, .75R: 0.792453, count: 53 Region 16 Avg IOU: 0.738812, Class: 0.999975, Obj: 0.843109, No Obj: 0.076680, .5R: 0.888889, .75R: 0.597222, count: 72 Region 16 Avg IOU: 0.739819, Class: 0.999935, Obj: 0.861053, No Obj: 0.073286, .5R: 0.942857, .75R: 0.600000, count: 70 Region 23 Avg IOU: 0.783096, Class: 0.999920, Obj: 0.808885, No Obj: 0.017017, .5R: 0.944444, .75R: 0.709877, count: 162 Region 23 Avg IOU: 0.700198, Class: 0.999966, Obj: 0.753024, No Obj: 0.008921, .5R: 0.831858, .75R: 0.495575, count: 113 Region 30 Avg IOU: 0.611063, Class: 0.999870, Obj: 0.344368, No Obj: 0.000286, .5R: 0.666667, .75R: 0.333333, count: 18 Region 16 Avg IOU: 0.747940, Class: 0.999982, Obj: 0.877716, No Obj: 0.017431, .5R: 0.812500, .75R: 0.625000, count: 16 Region 30 Avg IOU: 0.801591, Class: 0.999963, Obj: 0.794851, No Obj: 0.002891, .5R: 0.945652, .75R: 0.804348, count: 92 Region 23 Avg IOU: 0.817040, Class: 0.999910, Obj: 0.882889, No Obj: 0.014809, .5R: 0.969697, .75R: 0.810606, count: 132 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000328, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.811264, Class: 0.999982, Obj: 0.861043, No Obj: 0.003635, .5R: 1.000000, .75R: 0.807692, count: 26 Region 30 Avg IOU: 0.834934, Class: 0.999977, Obj: 0.898675, No Obj: 0.006247, .5R: 0.972527, .75R: 0.890110, count: 182 Region 16 Avg IOU: 0.707852, Class: 0.999944, Obj: 0.858346, No Obj: 0.098443, .5R: 0.903226, .75R: 0.526882, count: 93 Region 23 Avg IOU: 0.747902, Class: 0.999942, Obj: 0.732162, No Obj: 0.007036, .5R: 0.951219, .75R: 0.609756, count: 82 Region 30 Avg IOU: 0.786284, Class: 0.999992, Obj: 0.906183, No Obj: 0.010794, .5R: 0.951542, .75R: 0.746696, count: 454 Region 30 Avg IOU: 0.624224, Class: 0.999793, Obj: 0.468810, No Obj: 0.000325, .5R: 0.500000, .75R: 0.428571, count: 14 Region 16 Avg IOU: 0.823359, Class: 0.999987, Obj: 0.937686, No Obj: 0.033137, .5R: 1.000000, .75R: 0.769231, count: 26 Region 16 Avg IOU: 0.801340, Class: 0.999977, Obj: 0.755342, No Obj: 0.017258, .5R: 0.944444, .75R: 0.722222, count: 18 Region 23 Avg IOU: 0.818220, Class: 0.999974, Obj: 0.881579, No Obj: 0.015712, .5R: 0.954248, .75R: 0.849673, count: 153 Region 23 Avg IOU: 0.838390, Class: 0.999971, Obj: 0.915203, No Obj: 0.016098, .5R: 1.000000, .75R: 0.884615, count: 130 Region 30 Avg IOU: 0.831255, Class: 0.999963, Obj: 0.893451, No Obj: 0.003975, .5R: 0.990385, .75R: 0.826923, count: 104 Region 16 Avg IOU: 0.714124, Class: 0.999946, Obj: 0.858500, No Obj: 0.042607, .5R: 0.948718, .75R: 0.384615, count: 39 Region 23 Avg IOU: 0.702976, Class: 0.999946, Obj: 0.734318, No Obj: 0.012985, .5R: 0.885714, .75R: 0.480000, count: 175 Region 30 Avg IOU: 0.842396, Class: 0.999958, Obj: 0.905178, No Obj: 0.008607, .5R: 0.988550, .75R: 0.881679, count: 262 Region 16 Avg IOU: 0.768322, Class: 0.999997, Obj: 0.917121, No Obj: 0.001788, .5R: 1.000000, .75R: 1.000000, count: 1 Region 23 Avg IOU: 0.823302, Class: 0.999965, Obj: 0.918525, No Obj: 0.007879, .5R: 0.984127, .75R: 0.888889, count: 63 Region 30 Avg IOU: 0.695026, Class: 0.999994, Obj: 0.587920, No Obj: 0.000497, .5R: 0.785714, .75R: 0.571429, count: 14 Region 16 Avg IOU: 0.807127, Class: 0.999989, Obj: 0.903408, No Obj: 0.014255, .5R: 1.000000, .75R: 0.750000, count: 12 Region 23 Avg IOU: 0.831283, Class: 0.999972, Obj: 0.900378, No Obj: 0.013053, .5R: 0.981818, .75R: 0.900000, count: 110 Region 30 Avg IOU: 0.814325, Class: 0.999992, Obj: 0.899325, No Obj: 0.008392, .5R: 0.968153, .75R: 0.805732, count: 314 Region 16 Avg IOU: 0.819781, Class: 0.999989, Obj: 0.903841, No Obj: 0.034691, .5R: 1.000000, .75R: 0.884615, count: 26 Region 30 Avg IOU: 0.837198, Class: 0.999992, Obj: 0.912624, No Obj: 0.006308, .5R: 0.975369, .75R: 0.866995, count: 203 Region 16 Avg IOU: 0.761521, Class: 0.999985, Obj: 0.833169, No Obj: 0.041113, .5R: 0.925000, .75R: 0.675000, count: 40 Region 23 Avg IOU: 0.817798, Class: 0.999978, Obj: 0.931391, No Obj: 0.017814, .5R: 0.967949, .75R: 0.833333, count: 156 Region 23 Avg IOU: 0.804336, Class: 0.999967, Obj: 0.873768, No Obj: 0.015891, .5R: 0.979452, .75R: 0.787671, count: 146 Region 30 Avg IOU: 0.831505, Class: 0.999959, Obj: 0.855128, No Obj: 0.003824, .5R: 0.972727, .75R: 0.872727, count: 110 Region 30 Avg IOU: 0.841639, Class: 0.999935, Obj: 0.900327, No Obj: 0.006237, .5R: 0.994350, .75R: 0.841808, count: 177 Region 16 Avg IOU: 0.723505, Class: 0.999952, Obj: 0.846215, No Obj: 0.102748, .5R: 0.922330, .75R: 0.524272, count: 103 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.001620, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.772472, Class: 0.999922, Obj: 0.715603, No Obj: 0.007661, .5R: 0.903614, .75R: 0.698795, count: 83 Region 23 Avg IOU: 0.811732, Class: 0.999970, Obj: 0.865797, No Obj: 0.004987, .5R: 0.937500, .75R: 0.781250, count: 32 Region 30 Avg IOU: 0.653523, Class: 0.999835, Obj: 0.428542, No Obj: 0.000505, .5R: 0.642857, .75R: 0.571429, count: 14 Region 16 Avg IOU: 0.795611, Class: 0.999977, Obj: 0.846498, No Obj: 0.014837, .5R: 1.000000, .75R: 0.750000, count: 12 Region 23 Avg IOU: 0.756210, Class: 0.999974, Obj: 0.850234, No Obj: 0.014752, .5R: 0.916201, .75R: 0.675978, count: 179 Region 30 Avg IOU: 0.798154, Class: 0.999991, Obj: 0.903927, No Obj: 0.010909, .5R: 0.957589, .75R: 0.761161, count: 448 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000088, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 30 Avg IOU: 0.793542, Class: 0.999987, Obj: 0.849811, No Obj: 0.003215, .5R: 0.962264, .75R: 0.726415, count: 106 Region 23 Avg IOU: 0.807716, Class: 0.999967, Obj: 0.909803, No Obj: 0.005633, .5R: 0.942308, .75R: 0.846154, count: 52 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000556, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.728218, Class: 0.999957, Obj: 0.765718, No Obj: 0.001632, .5R: 0.818182, .75R: 0.454545, count: 11 Region 30 Avg IOU: 0.795240, Class: 0.999993, Obj: 0.879535, No Obj: 0.008378, .5R: 0.957187, .75R: 0.743119, count: 327 Region 16 Avg IOU: 0.804749, Class: 0.999978, Obj: 0.534415, No Obj: 0.000905, .5R: 1.000000, .75R: 1.000000, count: 1 Region 30 Avg IOU: 0.743168, Class: 0.999996, Obj: 0.859791, No Obj: 0.008270, .5R: 0.914667, .75R: 0.610667, count: 375 Region 23 Avg IOU: 0.835718, Class: 0.999976, Obj: 0.899011, No Obj: 0.003652, .5R: 1.000000, .75R: 0.923077, count: 26 Region 16 Avg IOU: 0.728175, Class: 0.999966, Obj: 0.808297, No Obj: 0.048727, .5R: 0.891304, .75R: 0.586957, count: 46 Region 23 Avg IOU: 0.730284, Class: 0.999971, Obj: 0.805616, No Obj: 0.010071, .5R: 0.916667, .75R: 0.522727, count: 132 Region 30 Avg IOU: 0.728481, Class: 0.999985, Obj: 0.625471, No Obj: 0.000746, .5R: 0.903226, .75R: 0.612903, count: 31 Region 16 Avg IOU: 0.756243, Class: 0.999973, Obj: 0.885283, No Obj: 0.066740, .5R: 0.888889, .75R: 0.730159, count: 63 Region 30 Avg IOU: 0.798812, Class: 0.999996, Obj: 0.867919, No Obj: 0.008740, .5R: 0.963788, .75R: 0.757660, count: 359 Region 23 Avg IOU: 0.818781, Class: 0.999962, Obj: 0.885960, No Obj: 0.013148, .5R: 0.974138, .75R: 0.844828, count: 116 Region 16 Avg IOU: 0.830917, Class: 0.999974, Obj: 0.963580, No Obj: 0.013290, .5R: 1.000000, .75R: 0.909091, count: 11 Region 23 Avg IOU: 0.834156, Class: 0.999974, Obj: 0.923455, No Obj: 0.013683, .5R: 0.983871, .75R: 0.846774, count: 124 Region 30 Avg IOU: 0.822619, Class: 0.999986, Obj: 0.846776, No Obj: 0.002133, .5R: 0.968750, .75R: 0.828125, count: 64 Region 16 Avg IOU: 0.735456, Class: 0.999920, Obj: 0.934690, No Obj: 0.003623, .5R: 1.000000, .75R: 0.666667, count: 3 Region 23 Avg IOU: 0.746354, Class: 0.999983, Obj: 0.839408, No Obj: 0.014395, .5R: 0.935135, .75R: 0.616216, count: 185 Region 30 Avg IOU: 0.825580, Class: 0.999987, Obj: 0.893646, No Obj: 0.007345, .5R: 0.974684, .75R: 0.848101, count: 237 Region 16 Avg IOU: 0.765672, Class: 0.999973, Obj: 0.825999, No Obj: 0.069791, .5R: 0.957143, .75R: 0.671429, count: 70 Region 23 Avg IOU: 0.785987, Class: 0.999898, Obj: 0.795942, No Obj: 0.015613, .5R: 0.940789, .75R: 0.756579, count: 152 Region 30 Avg IOU: 0.789481, Class: 0.999984, Obj: 0.826173, No Obj: 0.004271, .5R: 0.967742, .75R: 0.716129, count: 155 Region 16 Avg IOU: 0.868873, Class: 0.999998, Obj: 0.983086, No Obj: 0.008609, .5R: 1.000000, .75R: 1.000000, count: 3 Region 23 Avg IOU: 0.822496, Class: 0.999979, Obj: 0.897885, No Obj: 0.009936, .5R: 0.989362, .75R: 0.808511, count: 94 Region 30 Avg IOU: 0.813476, Class: 0.999960, Obj: 0.791598, No Obj: 0.002429, .5R: 0.955882, .75R: 0.794118, count: 68 Region 16 Avg IOU: 0.839227, Class: 0.999984, Obj: 0.977139, No Obj: 0.026227, .5R: 1.000000, .75R: 0.894737, count: 19 Region 23 Avg IOU: 0.819447, Class: 0.999970, Obj: 0.891195, No Obj: 0.011929, .5R: 0.959016, .75R: 0.827869, count: 122 Region 30 Avg IOU: 0.809976, Class: 0.999973, Obj: 0.868464, No Obj: 0.006420, .5R: 0.948357, .75R: 0.793427, count: 213 Region 16 Avg IOU: 0.811305, Class: 0.999956, Obj: 0.934606, No Obj: 0.010881, .5R: 1.000000, .75R: 0.714286, count: 7 Region 23 Avg IOU: 0.834152, Class: 0.999989, Obj: 0.925958, No Obj: 0.011167, .5R: 0.980000, .75R: 0.870000, count: 100 Region 30 Avg IOU: 0.829069, Class: 0.999992, Obj: 0.900042, No Obj: 0.004545, .5R: 0.968000, .75R: 0.880000, count: 125 Region 16 Avg IOU: 0.830712, Class: 0.999970, Obj: 0.962832, No Obj: 0.019166, .5R: 1.000000, .75R: 0.812500, count: 16 Region 23 Avg IOU: 0.759505, Class: 0.999970, Obj: 0.862703, No Obj: 0.017954, .5R: 0.924883, .75R: 0.685446, count: 213 Region 30 Avg IOU: 0.823933, Class: 0.999906, Obj: 0.906162, No Obj: 0.006840, .5R: 0.966346, .75R: 0.841346, count: 208 Region 16 Avg IOU: 0.828858, Class: 0.999945, Obj: 0.999447, No Obj: 0.002872, .5R: 1.000000, .75R: 1.000000, count: 1 Region 30 Avg IOU: 0.767867, Class: 0.999990, Obj: 0.831326, No Obj: 0.001834, .5R: 0.882353, .75R: 0.705882, count: 51 Region 16 Avg IOU: 0.683896, Class: 0.999997, Obj: 0.933464, No Obj: 0.002387, .5R: 1.000000, .75R: 0.000000, count: 1 Region 23 Avg IOU: 0.823557, Class: 0.999982, Obj: 0.911689, No Obj: 0.006363, .5R: 1.000000, .75R: 0.872340, count: 47 Region 23 Avg IOU: 0.739290, Class: 0.999989, Obj: 0.859175, No Obj: 0.013444, .5R: 0.895062, .75R: 0.641975, count: 162 Region 30 Avg IOU: 0.788183, Class: 0.999990, Obj: 0.826461, No Obj: 0.003887, .5R: 0.958042, .75R: 0.727273, count: 143 Region 30 Avg IOU: 0.821235, Class: 0.999995, Obj: 0.932197, No Obj: 0.008962, .5R: 0.964497, .75R: 0.828402, count: 338 Region 16 Avg IOU: 0.844340, Class: 0.999995, Obj: 0.743565, No Obj: 0.002285, .5R: 1.000000, .75R: 1.000000, count: 3 Region 16 Avg IOU: 0.686679, Class: 0.999951, Obj: 0.831300, No Obj: 0.081555, .5R: 0.835443, .75R: 0.481013, count: 79 Region 23 Avg IOU: 0.814791, Class: 0.999987, Obj: 0.917592, No Obj: 0.009837, .5R: 0.976471, .75R: 0.800000, count: 85 Region 23 Avg IOU: 0.756628, Class: 0.999937, Obj: 0.760754, No Obj: 0.009383, .5R: 0.973913, .75R: 0.652174, count: 115 Region 30 Avg IOU: 0.794926, Class: 0.999774, Obj: 0.671351, No Obj: 0.000859, .5R: 0.969697, .75R: 0.787879, count: 33 Region 16 Avg IOU: 0.734111, Class: 0.999964, Obj: 0.865978, No Obj: 0.083815, .5R: 0.930233, .75R: 0.546512, count: 86 Region 30 Avg IOU: 0.826153, Class: 0.999995, Obj: 0.924704, No Obj: 0.006579, .5R: 0.963801, .75R: 0.859729, count: 221 Region 23 Avg IOU: 0.786931, Class: 0.999970, Obj: 0.848437, No Obj: 0.011661, .5R: 0.966102, .75R: 0.720339, count: 118 Region 16 Avg IOU: 0.739137, Class: 0.999966, Obj: 0.835421, No Obj: 0.044483, .5R: 0.894737, .75R: 0.526316, count: 38 Region 23 Avg IOU: 0.787825, Class: 0.999905, Obj: 0.826967, No Obj: 0.013027, .5R: 0.978417, .75R: 0.741007, count: 139 Region 30 Avg IOU: 0.708966, Class: 0.999966, Obj: 0.688849, No Obj: 0.001135, .5R: 0.805556, .75R: 0.527778, count: 36 Region 16 Avg IOU: 0.687439, Class: 0.999965, Obj: 0.905719, No Obj: 0.004698, .5R: 1.000000, .75R: 0.333333, count: 3 Region 30 Avg IOU: 0.793689, Class: 0.999981, Obj: 0.826000, No Obj: 0.002307, .5R: 0.920000, .75R: 0.773333, count: 75 Region 16 Avg IOU: 0.695431, Class: 0.999961, Obj: 0.779716, No Obj: 0.089711, .5R: 0.851064, .75R: 0.478723, count: 94 Region 23 Avg IOU: 0.808466, Class: 0.999976, Obj: 0.919033, No Obj: 0.008497, .5R: 0.987342, .75R: 0.835443, count: 79 Region 23 Avg IOU: 0.744129, Class: 0.999778, Obj: 0.707510, No Obj: 0.010251, .5R: 0.937500, .75R: 0.633929, count: 112 Region 30 Avg IOU: 0.762583, Class: 0.999985, Obj: 0.628209, No Obj: 0.000842, .5R: 0.964286, .75R: 0.642857, count: 28 Region 16 Avg IOU: 0.817415, Class: 0.999990, Obj: 0.935950, No Obj: 0.024665, .5R: 1.000000, .75R: 0.842105, count: 19 Region 30 Avg IOU: 0.801141, Class: 0.999991, Obj: 0.891405, No Obj: 0.006743, .5R: 0.973485, .75R: 0.776515, count: 264 Region 16 Avg IOU: 0.809184, Class: 0.999970, Obj: 0.876042, No Obj: 0.029681, .5R: 1.000000, .75R: 0.695652, count: 23 Region 23 Avg IOU: 0.828909, Class: 0.999964, Obj: 0.894897, No Obj: 0.018004, .5R: 0.968553, .75R: 0.849057, count: 159 Region 23 Avg IOU: 0.821317, Class: 0.999967, Obj: 0.904160, No Obj: 0.015156, .5R: 0.964789, .75R: 0.823944, count: 142 Region 30 Avg IOU: 0.841089, Class: 0.999968, Obj: 0.892012, No Obj: 0.007526, .5R: 0.981481, .75R: 0.870370, count: 216 Region 30 Avg IOU: 0.840739, Class: 0.999976, Obj: 0.888457, No Obj: 0.005702, .5R: 0.993976, .75R: 0.867470, count: 166 Region 16 Avg IOU: 0.781864, Class: 0.999953, Obj: 0.881708, No Obj: 0.041029, .5R: 0.972222, .75R: 0.666667, count: 36 Region 16 Avg IOU: 0.764376, Class: 0.999987, Obj: 0.878910, No Obj: 0.051864, .5R: 0.913043, .75R: 0.652174, count: 46 Region 23 Avg IOU: 0.751753, Class: 0.999971, Obj: 0.836366, No Obj: 0.012956, .5R: 0.949045, .75R: 0.611465, count: 157 Region 23 Avg IOU: 0.804121, Class: 0.999928, Obj: 0.861406, No Obj: 0.017627, .5R: 0.969697, .75R: 0.763636, count: 165 Region 30 Avg IOU: 0.677028, Class: 0.999953, Obj: 0.525213, No Obj: 0.000462, .5R: 0.750000, .75R: 0.562500, count: 16 Region 16 Avg IOU: 0.782719, Class: 0.999969, Obj: 0.899336, No Obj: 0.044243, .5R: 0.953488, .75R: 0.651163, count: 43 Region 23 Avg IOU: 0.753807, Class: 0.999974, Obj: 0.851542, No Obj: 0.014725, .5R: 0.925714, .75R: 0.645714, count: 175 Region 30 Avg IOU: 0.834854, Class: 0.999956, Obj: 0.849065, No Obj: 0.003577, .5R: 0.990000, .75R: 0.820000, count: 100 Region 16 Avg IOU: 0.749831, Class: 0.999984, Obj: 0.838192, No Obj: 0.050088, .5R: 0.956522, .75R: 0.478261, count: 46 Region 23 Avg IOU: 0.814223, Class: 0.999950, Obj: 0.866806, No Obj: 0.020239, .5R: 0.978261, .75R: 0.815217, count: 184 Region 30 Avg IOU: 0.721071, Class: 0.999985, Obj: 0.644253, No Obj: 0.000839, .5R: 0.826087, .75R: 0.565217, count: 23 Region 16 Avg IOU: 0.784772, Class: 0.999982, Obj: 0.901517, No Obj: 0.016070, .5R: 0.928571, .75R: 0.785714, count: 14 Region 23 Avg IOU: 0.827615, Class: 0.999980, Obj: 0.913604, No Obj: 0.014455, .5R: 0.984496, .75R: 0.813953, count: 129 Region 30 Avg IOU: 0.833156, Class: 0.999951, Obj: 0.862914, No Obj: 0.005820, .5R: 0.975904, .75R: 0.897590, count: 166 Region 16 Avg IOU: 0.726489, Class: 0.999987, Obj: 0.897475, No Obj: 0.012095, .5R: 0.833333, .75R: 0.583333, count: 12 Region 23 Avg IOU: 0.837501, Class: 0.999979, Obj: 0.910322, No Obj: 0.009851, .5R: 1.000000, .75R: 0.895349, count: 86 Region 30 Avg IOU: 0.844345, Class: 0.999991, Obj: 0.913254, No Obj: 0.006977, .5R: 1.000000, .75R: 0.883117, count: 231 Region 16 Avg IOU: 0.823989, Class: 0.999997, Obj: 0.723179, No Obj: 0.003171, .5R: 1.000000, .75R: 1.000000, count: 3 Region 23 Avg IOU: 0.810352, Class: 0.999977, Obj: 0.880109, No Obj: 0.008153, .5R: 0.985294, .75R: 0.779412, count: 68 Region 30 Avg IOU: 0.838214, Class: 0.999997, Obj: 0.923439, No Obj: 0.004910, .5R: 0.968153, .75R: 0.878981, count: 157 Region 16 Avg IOU: 0.855575, Class: 0.999993, Obj: 0.998527, No Obj: 0.003324, .5R: 1.000000, .75R: 1.000000, count: 2 Region 23 Avg IOU: 0.841747, Class: 0.999961, Obj: 0.896440, No Obj: 0.007184, .5R: 1.000000, .75R: 0.931035, count: 58 Region 30 Avg IOU: 0.800070, Class: 0.999987, Obj: 0.869960, No Obj: 0.010351, .5R: 0.947090, .75R: 0.748677, count: 378 Region 16 Avg IOU: 0.851512, Class: 0.999986, Obj: 0.990764, No Obj: 0.007276, .5R: 1.000000, .75R: 0.833333, count: 6 Region 23 Avg IOU: 0.810471, Class: 0.999994, Obj: 0.928741, No Obj: 0.009660, .5R: 0.965909, .75R: 0.829545, count: 88 Region 30 Avg IOU: 0.820121, Class: 0.999990, Obj: 0.913992, No Obj: 0.010407, .5R: 0.980978, .75R: 0.812500, count: 368 Region 16 Avg IOU: 0.715895, Class: 0.999975, Obj: 0.795910, No Obj: 0.030892, .5R: 0.838710, .75R: 0.483871, count: 31 Region 23 Avg IOU: 0.748909, Class: 0.999960, Obj: 0.819991, No Obj: 0.014581, .5R: 0.932203, .75R: 0.638418, count: 177 Region 30 Avg IOU: 0.830100, Class: 0.999993, Obj: 0.908200, No Obj: 0.007252, .5R: 0.982833, .75R: 0.858369, count: 233 Region 16 Avg IOU: 0.764322, Class: 0.999986, Obj: 0.894828, No Obj: 0.022776, .5R: 0.952381, .75R: 0.666667, count: 21 Region 23 Avg IOU: 0.840994, Class: 0.999973, Obj: 0.921263, No Obj: 0.015174, .5R: 0.984375, .75R: 0.890625, count: 128 Region 30 Avg IOU: 0.800070, Class: 0.999990, Obj: 0.769804, No Obj: 0.002574, .5R: 0.980198, .75R: 0.762376, count: 101 Region 16 Avg IOU: 0.829293, Class: 0.999980, Obj: 0.932202, No Obj: 0.008098, .5R: 1.000000, .75R: 1.000000, count: 4 Region 23 Avg IOU: 0.784154, Class: 0.999987, Obj: 0.886683, No Obj: 0.014075, .5R: 0.947368, .75R: 0.743421, count: 152 Region 30 Avg IOU: 0.836122, Class: 0.999989, Obj: 0.900905, No Obj: 0.005857, .5R: 0.958824, .75R: 0.858824, count: 170 Region 16 Avg IOU: 0.786238, Class: 0.999982, Obj: 0.903853, No Obj: 0.013818, .5R: 1.000000, .75R: 0.727273, count: 11 Region 23 Avg IOU: 0.847247, Class: 0.999983, Obj: 0.929622, No Obj: 0.011979, .5R: 1.000000, .75R: 0.920455, count: 88 Region 30 Avg IOU: 0.784996, Class: 0.999989, Obj: 0.840831, No Obj: 0.004312, .5R: 0.949367, .75R: 0.702532, count: 158 Region 16 Avg IOU: 0.821757, Class: 0.999986, Obj: 0.856214, No Obj: 0.016934, .5R: 1.000000, .75R: 0.769231, count: 13 Region 23 Avg IOU: 0.833958, Class: 0.999971, Obj: 0.920705, No Obj: 0.015500, .5R: 0.969925, .75R: 0.887218, count: 133 Region 30 Avg IOU: 0.836080, Class: 0.999994, Obj: 0.911600, No Obj: 0.005904, .5R: 0.975000, .75R: 0.880000, count: 200 Region 16 Avg IOU: 0.860321, Class: 0.999982, Obj: 0.943921, No Obj: 0.013424, .5R: 1.000000, .75R: 0.909091, count: 11 Region 23 Avg IOU: 0.846539, Class: 0.999977, Obj: 0.907655, No Obj: 0.012140, .5R: 1.000000, .75R: 0.897196, count: 107 Region 30 Avg IOU: 0.846313, Class: 0.999963, Obj: 0.919429, No Obj: 0.010122, .5R: 0.974922, .75R: 0.890282, count: 319 Region 16 Avg IOU: 0.804526, Class: 0.999985, Obj: 0.883302, No Obj: 0.014614, .5R: 1.000000, .75R: 0.785714, count: 14 Region 30 Avg IOU: 0.827951, Class: 0.999989, Obj: 0.913083, No Obj: 0.006716, .5R: 0.970874, .75R: 0.844660, count: 206 Region 23 Avg IOU: 0.831154, Class: 0.999964, Obj: 0.906198, No Obj: 0.014267, .5R: 0.958333, .75R: 0.866667, count: 120 Region 16 Avg IOU: 0.699172, Class: 0.999980, Obj: 0.731805, No Obj: 0.010880, .5R: 0.900000, .75R: 0.200000, count: 10 Region 23 Avg IOU: 0.761019, Class: 0.999982, Obj: 0.804587, No Obj: 0.009781, .5R: 0.946903, .75R: 0.672566, count: 113 Region 30 Avg IOU: 0.851578, Class: 0.999990, Obj: 0.925308, No Obj: 0.007755, .5R: 0.995851, .75R: 0.921162, count: 241 Region 30 Avg IOU: 0.780085, Class: 0.999993, Obj: 0.804418, No Obj: 0.004507, .5R: 0.962162, .75R: 0.724324, count: 185 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000318, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 16 Avg IOU: 0.616816, Class: 0.999960, Obj: 0.959947, No Obj: 0.004170, .5R: 1.000000, .75R: 0.000000, count: 1 Region 23 Avg IOU: 0.819485, Class: 0.999991, Obj: 0.907204, No Obj: 0.004251, .5R: 1.000000, .75R: 0.766667, count: 30 Region 23 Avg IOU: 0.820923, Class: 0.999995, Obj: 0.885072, No Obj: 0.004653, .5R: 1.000000, .75R: 0.790698, count: 43 Region 30 Avg IOU: 0.799465, Class: 0.999995, Obj: 0.878893, No Obj: 0.008053, .5R: 0.979290, .75R: 0.751479, count: 338 Region 30 Avg IOU: 0.807331, Class: 0.999991, Obj: 0.915093, No Obj: 0.010080, .5R: 0.979221, .75R: 0.789610, count: 385 Region 16 Avg IOU: 0.789021, Class: 0.999979, Obj: 0.922653, No Obj: 0.019710, .5R: 0.937500, .75R: 0.812500, count: 16 Region 16 Avg IOU: 0.679676, Class: 0.999958, Obj: 0.769796, No Obj: 0.018956, .5R: 0.842105, .75R: 0.315789, count: 19 Region 23 Avg IOU: 0.695239, Class: 0.999983, Obj: 0.775054, No Obj: 0.009923, .5R: 0.898551, .75R: 0.398551, count: 138 Region 23 Avg IOU: 0.810022, Class: 0.999976, Obj: 0.866698, No Obj: 0.013862, .5R: 0.969231, .75R: 0.807692, count: 130 Region 30 Avg IOU: 0.734211, Class: 0.999998, Obj: 0.642476, No Obj: 0.000749, .5R: 0.939394, .75R: 0.545455, count: 33 Region 16 Avg IOU: 0.767734, Class: 0.999976, Obj: 0.833824, No Obj: 0.023392, .5R: 0.954545, .75R: 0.727273, count: 22 Region 23 Avg IOU: 0.802979, Class: 0.999989, Obj: 0.883146, No Obj: 0.011415, .5R: 0.950000, .75R: 0.770000, count: 100 Region 30 Avg IOU: 0.837209, Class: 0.999987, Obj: 0.903611, No Obj: 0.007106, .5R: 0.973958, .75R: 0.869792, count: 192 Region 16 Avg IOU: 0.668998, Class: 0.999981, Obj: 0.987828, No Obj: 0.010998, .5R: 0.750000, .75R: 0.500000, count: 8 Region 23 Avg IOU: 0.827531, Class: 0.999971, Obj: 0.911575, No Obj: 0.012374, .5R: 0.974138, .75R: 0.844828, count: 116 Region 30 Avg IOU: 0.820669, Class: 0.999988, Obj: 0.896960, No Obj: 0.003688, .5R: 0.981308, .75R: 0.869159, count: 107 Region 16 Avg IOU: 0.806319, Class: 0.999984, Obj: 0.971032, No Obj: 0.024281, .5R: 1.000000, .75R: 0.684211, count: 19 Region 23 Avg IOU: 0.812562, Class: 0.999981, Obj: 0.916872, No Obj: 0.010771, .5R: 0.980198, .75R: 0.762376, count: 101 Region 30 Avg IOU: 0.841251, Class: 0.999986, Obj: 0.909663, No Obj: 0.007547, .5R: 0.996109, .75R: 0.887160, count: 257 Region 30 Avg IOU: 0.808331, Class: 0.999990, Obj: 0.874044, No Obj: 0.003221, .5R: 0.941176, .75R: 0.803922, count: 102 Region 16 Avg IOU: 0.810921, Class: 0.999983, Obj: 0.711703, No Obj: 0.014809, .5R: 1.000000, .75R: 0.800000, count: 15 Region 16 Avg IOU: 0.731560, Class: 0.999985, Obj: 0.856055, No Obj: 0.040721, .5R: 0.882353, .75R: 0.647059, count: 34 Region 23 Avg IOU: 0.826202, Class: 0.999980, Obj: 0.916940, No Obj: 0.013367, .5R: 0.973913, .75R: 0.860870, count: 115 Region 23 Avg IOU: 0.822452, Class: 0.999928, Obj: 0.852402, No Obj: 0.019589, .5R: 0.983784, .75R: 0.837838, count: 185 Region 30 Avg IOU: 0.825854, Class: 0.999987, Obj: 0.906565, No Obj: 0.006243, .5R: 0.972222, .75R: 0.842593, count: 216 Region 16 Avg IOU: 0.814519, Class: 0.999987, Obj: 0.933728, No Obj: 0.002511, .5R: 1.000000, .75R: 1.000000, count: 2 Region 30 Avg IOU: 0.839198, Class: 0.999977, Obj: 0.873315, No Obj: 0.005345, .5R: 0.973856, .75R: 0.882353, count: 153 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.002313, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.832005, Class: 0.999976, Obj: 0.895711, No Obj: 0.006831, .5R: 1.000000, .75R: 0.925926, count: 54 Region 23 Avg IOU: 0.828190, Class: 0.999992, Obj: 0.938989, No Obj: 0.006300, .5R: 1.000000, .75R: 0.934783, count: 46 Region 30 Avg IOU: 0.819601, Class: 0.999992, Obj: 0.909005, No Obj: 0.008410, .5R: 0.968847, .75R: 0.819315, count: 321 Region 16 Avg IOU: 0.776401, Class: 0.999954, Obj: 0.868690, No Obj: 0.011609, .5R: 1.000000, .75R: 0.800000, count: 10 Region 30 Avg IOU: 0.816808, Class: 0.999992, Obj: 0.920379, No Obj: 0.009651, .5R: 0.964481, .75R: 0.833333, count: 366 Region 23 Avg IOU: 0.753238, Class: 0.999982, Obj: 0.858025, No Obj: 0.015354, .5R: 0.914894, .75R: 0.659574, count: 188 Region 16 Avg IOU: 0.807330, Class: 0.999975, Obj: 0.864445, No Obj: 0.015280, .5R: 1.000000, .75R: 0.777778, count: 9 Region 23 Avg IOU: 0.798456, Class: 0.999975, Obj: 0.930415, No Obj: 0.011012, .5R: 0.954955, .75R: 0.810811, count: 111 Region 30 Avg IOU: 0.804671, Class: 0.999982, Obj: 0.837926, No Obj: 0.004253, .5R: 0.972789, .75R: 0.775510, count: 147 Region 30 Avg IOU: 0.831704, Class: 0.999980, Obj: 0.867492, No Obj: 0.005674, .5R: 0.985148, .75R: 0.856436, count: 202 Region 16 Avg IOU: -nan(ind), Class: -nan(ind), Obj: -nan(ind), No Obj: 0.000009, .5R: -nan(ind), .75R: -nan(ind), count: 0 Region 23 Avg IOU: 0.832461, Class: 0.999994, Obj: 0.798156, No Obj: 0.001867, .5R: 1.000000, .75R: 0.818182, count: 11 Region 30 Avg IOU: 0.752878, Class: 0.999991, Obj: 0.838917, No Obj: 0.008637, .5R: 0.937811, .75R: 0.609453, count: 402 Syncing... Done!

(next mAP calculation at 8040 iterations) Last accuracy mAP@0.5 = 23.18 % 8004: 29.937290, 28.569744 avg loss, 0.000200 rate, 31.780000 seconds, 1024512 images

E:\trainingdata\baochuang\test\newtest\3-14all\train>darknet detector map baocuang.data yolov3-tiny_3l_1.cfg yolov3-tiny_3l_1_final.weights compute_capability = 610, cudnn_half = 0 layer filters size input output 0 conv 16 3 x 3 / 1 1024 x1024 x 3 -> 1024 x1024 x 16 0.906 BF 1 max 2 x 2 / 2 1024 x1024 x 16 -> 512 x 512 x 16 0.017 BF 2 conv 32 3 x 3 / 1 512 x 512 x 16 -> 512 x 512 x 32 2.416 BF 3 max 2 x 2 / 2 512 x 512 x 32 -> 256 x 256 x 32 0.008 BF 4 conv 64 3 x 3 / 1 256 x 256 x 32 -> 256 x 256 x 64 2.416 BF 5 max 2 x 2 / 2 256 x 256 x 64 -> 128 x 128 x 64 0.004 BF 6 conv 128 3 x 3 / 1 128 x 128 x 64 -> 128 x 128 x 128 2.416 BF 7 max 2 x 2 / 2 128 x 128 x 128 -> 64 x 64 x 128 0.002 BF 8 conv 256 3 x 3 / 1 64 x 64 x 128 -> 64 x 64 x 256 2.416 BF 9 max 2 x 2 / 2 64 x 64 x 256 -> 32 x 32 x 256 0.001 BF 10 conv 512 3 x 3 / 1 32 x 32 x 256 -> 32 x 32 x 512 2.416 BF 11 max 2 x 2 / 1 32 x 32 x 512 -> 32 x 32 x 512 0.002 BF 12 conv 1024 3 x 3 / 1 32 x 32 x 512 -> 32 x 32 x1024 9.664 BF 13 conv 256 1 x 1 / 1 32 x 32 x1024 -> 32 x 32 x 256 0.537 BF 14 conv 512 3 x 3 / 1 32 x 32 x 256 -> 32 x 32 x 512 2.416 BF 15 conv 8 1 x 1 / 1 32 x 32 x 512 -> 32 x 32 x 8 0.008 BF 16 yolo 17 route 13 18 conv 128 1 x 1 / 1 32 x 32 x 256 -> 32 x 32 x 128 0.067 BF 19 upsample 2x 32 x 32 x 128 -> 64 x 64 x 128 20 route 19 8 21 conv 256 3 x 3 / 1 64 x 64 x 384 -> 64 x 64 x 256 7.248 BF 22 conv 32 1 x 1 / 1 64 x 64 x 256 -> 64 x 64 x 32 0.067 BF 23 yolo 24 route 21 25 conv 128 1 x 1 / 1 64 x 64 x 256 -> 64 x 64 x 128 0.268 BF 26 upsample 2x 64 x 64 x 128 -> 128 x 128 x 128 27 route 26 6 28 conv 128 3 x 3 / 1 128 x 128 x 256 -> 128 x 128 x 128 9.664 BF 29 conv 32 1 x 1 / 1 128 x 128 x 128 -> 128 x 128 x 32 0.134 BF 30 yolo Total BFLOPS 43.093 Allocate additional workspace_size = 52.43 MB Loading weights from yolov3-tiny_3l_1_final.weights... seen 64 Done!

calculation mAP (mean average precision)... 3 detections_count = 1727, unique_truth_count = 1204 class_id = 0, name = sperm, ap = 69.15% (TP = 956, FP = 310) class_id = 1, name = round cell, ap = 0.00% (TP = 0, FP = 0) class_id = 2, name = red cell, ap = 0.00% (TP = 0, FP = 0)

for thresh = 0.25, precision = 0.76, recall = 0.79, F1-score = 0.77 for thresh = 0.25, TP = 956, FP = 310, FN = 248, average IoU = 49.44 %

IoU threshold = 50 %, used Area-Under-Curve for each unique Recall mean average precision (mAP@0.50) = 0.230516, or 23.05 % Total Detection Time: 0.000000 Seconds

Set -points flag: -points 101 for MS COCO -points 11 for PascalVOC 2007 (uncomment difficult in voc.data) -points 0 (AUC) for ImageNet, PascalVOC 2010-2012, your custom dataset

AlexeyAB commented 5 years ago

@wting861006

class_id = 0, name = sperm, ap = 69.15% (TP = 956, FP = 310)

mAP ~ 70% is a good result for Yolov3 Tiny 3L.