Hi, I think that I’ve tried everything I could, so I’d be glad if someone could help me. Also I didn’t find the answer in the Internet or the forums, it seems like a bug for me, because it should just work.
I’m trying just to add new labels to pretrained dataset, I’ve made many tries and got to the place, where simplest change in yaml is getting me nowhere, so:
My train1.py file:
from ultralytics import YOLO
model = YOLO(‘yolov5n.pt’)
results = model.train(data=‘coco2017_custom.yaml’, epochs=1, pretrained=True)results
I’ve downloaded yolov5/data/coco128.yaml at master · ultralytics/yolov5 · GitHub (which is often refered in tutorials), saved to my workspace directory as coco2017_custom.yaml, ran training and everything works well:
YOLOv5n summary (fused): 193 layers, 2649200 parameters, 0 gradients, 7.7 GFLOPs
Class Images Instances Box(P R mAP50 mAP50-95): 100%|██████████| 4/4 [00:18<00
all 128 929 0.658 0.49 0.578 0.419
person 128 254 0.787 0.673 0.759 0.505
bicycle 128 6 0.525 0.192 0.48 0.255
(and so on)
BUT when I add just one class to the dowloaded yaml file:
# YOLOv5 🚀 by Ultralytics, AGPL-3.0 license
# COCO128 dataset https://www.kaggle.com/ultralytics/coco128 (first 128 images from COCO train2017) by Ultralytics
# Example usage: python train.py --data coco128.yaml
# parent
# ├── yolov5
# └── datasets
# └── coco128 ← downloads here (7 MB)
# Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]
path: ../datasets/coco128 # dataset root dir
train: images/train2017 # train images (relative to 'path') 128 images
val: images/train2017 # val images (relative to 'path') 128 images
test: # test images (optional)
# Classes
names:
0: person
1: bicycle
2: car
3: motorcycle
4: airplane
5: bus
6: train
7: truck
8: boat
9: traffic light
10: fire hydrant
11: stop sign
12: parking meter
13: bench
14: bird
15: cat
16: dog
17: horse
18: sheep
19: cow
20: elephant
21: bear
22: zebra
23: giraffe
24: backpack
25: umbrella
26: handbag
27: tie
28: suitcase
29: frisbee
30: skis
31: snowboard
32: sports ball
33: kite
34: baseball bat
35: baseball glove
36: skateboard
37: surfboard
38: tennis racket
39: bottle
40: wine glass
41: cup
42: fork
43: knife
44: spoon
45: bowl
46: banana
47: apple
48: sandwich
49: orange
50: broccoli
51: carrot
52: hot dog
53: pizza
54: donut
55: cake
56: chair
57: couch
58: potted plant
59: bed
60: dining table
61: toilet
62: tv
63: laptop
64: mouse
65: remote
66: keyboard
67: cell phone
68: microwave
69: oven
70: toaster
71: sink
72: refrigerator
73: book
74: clock
75: vase
76: scissors
77: teddy bear
78: hair drier
79: toothbrush
80: xxx
# Download script/URL (optional)
download: https://ultralytics.com/assets/coco128.zip
(please note last entry in names field), run test, then NONE of labels are detected by the training. (I use the same train1.py as before):
Validating runs/detect/train/weights/best.pt...
Ultralytics YOLOv8.0.83 🚀 Python-3.10.8 torch-1.13.1 CPU
YOLOv5n summary (fused): 193 layers, 2658071 parameters, 0 gradients, 7.8 GFLOPs
Class Images Instances Box(P R mAP50 mAP50-95): 100%|██████████| 4/4 [00:18<00
all 128 929 0 0 0 0
(and thats all the labels in the summary, no other which were present in the previous example)
Also using this model in inference produces NO DETECTIONS, while original, trained on unchanged yaml file (but loaded from my workspace local directory) is working well.
I didn’t change anything in train data at to this point, but later I’ve edited one of label files just to include new label in datasets, and it produces the same results. With added label to the yaml file, the training script is detecting change in config:
Overriding model.yaml nc=80 with nc=81
I’ve tried this using yolov5n.pt and yolov8n.pt pretrained models, the results are the same.
Edit: some typos and clarifications