search

mystic123 / tensorflow-yolo-v3

Implementation of YOLO v3 object detector in Tensorflow (TF-Slim)
https://medium.com/@pawekapica_31302/implementing-yolo-v3-in-tensorflow-tf-slim-c3c55ff59dbe
Apache License 2.0
894 stars 353 forks source link

mAp is less 0.2 than darknet. Could you help me ? #80

Open lucy3589 opened 5 years ago

lucy3589 commented 5 years ago

mAp is less 0.2 than darknet. Could you help me ?My address is 772152438@qq.com

atsunori commented 4 years ago

I measured mAP using this model (mystic123/tensorflow-yolo-v3). Measurement was performed by remodeling validation_app of OpenVINO. However, the performance value is different from the paper. Is there a problem with this model? Please let me know if you notice anything.

mAP measurement conditions: YOLOv3-416@IoU=0.5

OpenVINO validation_app result: [ INFO ] InferenceEngine: API version ............ 1.6 Build .................. custom_releases/2019/R1_c9b66a26e4d65bb986bb740e73f58c6e9e84c7c2 [ INFO ] Parsing input parameters [ INFO ] Loading plugin

API version ............ 1.6
Build .................. 22443
Description ....... MKLDNNPlugin

[ INFO ] Loading network files [ INFO ] Preparing input blobs [ INFO ] Batch size is 1 [ INFO ] Device: CPU [ INFO ] Collecting VOC annotations from /home/dla/sumi/coco/annotations_pascalformat [ INFO ] 5000 annotations collected [ INFO ] Starting inference Progress: [....................] 100.00% done [ INFO ] Processing output blobs [ INFO ] Inference report: Network load time: 112.53ms Model: mo/yolo_v3.xml Model Precision: FP32 Batch size: 1 Validation dataset: /home/dla/sumi Validation approach: Object detection network [ INFO ] Average infer time (ms): 280.48 (3.56532655 images per second with batch size = 1) Average precision per class table:

Class AP 1 0.329 2 0.268 3 0.173 4 0.426 5 0.618 6 0.618 7 0.812 8 0.355 9 0.213 10 0.091 11 0.453 12 0.452 13 0.347 14 0.320 15 0.210 16 0.848 17 0.632 18 0.492 19 0.301 20 0.291 21 0.582 22 0.729 23 0.543 24 0.632 25 0.117 26 0.350 27 0.099 28 0.179 29 0.297 30 0.182 31 0.197 32 0.257 33 0.091 34 0.157 35 0.258 36 0.091 37 0.356 38 0.312 39 0.312 40 0.091 41 0.165 42 0.174 43 0.305 44 0.159 45 0.130 46 0.336 47 0.191 48 0.122 49 0.310 50 0.151 51 0.212 52 0.091 53 0.216 54 0.410 55 0.166 56 0.323 57 0.211 58 0.671 59 0.215 60 0.706 61 0.460 62 0.690 63 0.504 64 0.620 65 0.165 66 0.091 67 0.575 68 0.208 69 0.403 70 0.510 71 0.036 72 0.378 73 0.565 74 0.091 75 0.264 76 0.219 77 0.429 78 0.481 79 0.051 80 0.169

Mean Average Precision (mAP): 0.3282