search

NVIDIA / DeepLearningExamples

State-of-the-Art Deep Learning scripts organized by models - easy to train and deploy with reproducible accuracy and performance on enterprise-grade infrastructure.
13.59k stars 3.24k forks source link

[SSD/PyTorch] The JSON element does not have the requested type #1422

Open ksv87 opened 2 months ago

ksv87 commented 2 months ago

Related to Model/Framework(s) SSD/PyTorch

Describe the bug

DLL 2024-09-05 08:00:14.587129 - PARAMETER dataset path : /coco  epochs : 65  batch size : 64│F1Help  F2Setup F3SearchF4FilterF5List  F6SortByF7Nice -F8Nice +F9Kill  F10Quit
  eval batch size : 32  no cuda : False  seed : None  checkpoint path : None  mode : training├──────────────────────────────────────────────────────────────────────────────────────────────  eval on epochs : [21, 31, 37, 42, 48, 53, 59, 64]  lr decay epochs : [43, 54]  learning rat│ksv@neroksv:~$
e : 0.0026  momentum : 0.9  weight decay : 0.0005  lr warmup : 300  backbone : resnet50  back│
bone path : None  num workers : 8  AMP : True  precision : amp                               │
Using seed = 6772                                                                            │
Loading annotations into memory...                                                           │
Done (t=0.03s)                                                                               │
Creating index...                                                                            │
terminate called after throwing an instance of 'simdjson::simdjson_error'                    │
  what():  The JSON element does not have the requested type.                                │
Aborted (core dumped)

To Reproduce Steps to reproduce the behavior:

  1. Install https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/Detection/SSD#quick-start-guide
  2. Set COCO_DIR to my custom dataset
  3. Change https://github.com/NVIDIA/DeepLearningExamples/blob/729963dd47e7c8bd462ad10bfac7a7b0b604e6dd/PyTorch/Detection/SSD/ssd/model.py#L61C9-L61C54 to my class count + 1
  4. Launch python main.py --backbone resnet50 --warmup 300 --bs 64 --d│ 565 root 0.0 0.0 0:00.58 ├─ /lib/systemd/systemd-journald ata /coco

Expected behavior A clear and concise description of what you expected to happen.

Environment Please provide at least: