Object detection API mAP values ​​are 0 although my training data is good

[Edi2410]

1. The entire URL of the file you are using

https://github.com/tensorflow/models/tree/master/research/object_detection

2. Describe the bug

It happens to me that the data during training (loss) is very good, even too good in one part, but I accept that because the pictures are very similar. Unfortunately, what happens to me is that during the evaluation, all my mAP and other data are equal to 0.

WARNING:tensorflow:Forced number of epochs for all eval validations to be 1. W0912 15:59:00.009776 140088353853824 model_lib_v2.py:1089] Forced number of epochs for all eval validations to be 1. INFO:tensorflow:Maybe overwriting sample_1_of_n_eval_examples: None I0912 15:59:00.009941 140088353853824 config_util.py:552] Maybe overwriting sample_1_of_n_eval_examples: None INFO:tensorflow:Maybe overwriting use_bfloat16: False I0912 15:59:00.010002 140088353853824 config_util.py:552] Maybe overwriting use_bfloat16: False INFO:tensorflow:Maybe overwriting eval_num_epochs: 1 I0912 15:59:00.010061 140088353853824 config_util.py:552] Maybe overwriting eval_num_epochs: 1 WARNING:tensorflow:Expected number of evaluation epochs is 1, but instead encountered eval_on_train_input_config.num_epochs = 0. Overwriting num_epochs to 1. W0912 15:59:00.010136 140088353853824 model_lib_v2.py:1106] Expected number of evaluation epochs is 1, but instead encountered eval_on_train_input_config.num_epochs = 0. Overwriting num_epochs to 1. 2022-09-12 15:59:00.028241: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 FMA To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags. 2022-09-12 15:59:00.962258: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1532] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 7405 MB memory: -> device: 0, name: NVIDIA GeForce GTX 1070, pci bus id: 0000:01:00.0, compute capability: 6.1 2022-09-12 15:59:00.963072: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1532] Created device /job:localhost/replica:0/task:0/device:GPU:1 with 7078 MB memory: -> device: 1, name: NVIDIA GeForce GTX 1070, pci bus id: 0000:02:00.0, compute capability: 6.1 INFO:tensorflow:Reading unweighted datasets: ['./training_outlook_action_ctx/data/val.records'] I0912 15:59:01.027812 140088353853824 dataset_builder.py:162] Reading unweighted datasets: ['./training_outlook_action_ctx/data/val.records'] INFO:tensorflow:Reading record datasets for input file: ['./training_outlook_action_ctx/data/val.records'] I0912 15:59:01.027991 140088353853824 dataset_builder.py:79] Reading record datasets for input file: ['./training_outlook_action_ctx/data/val.records'] INFO:tensorflow:Number of filenames to read: 1 I0912 15:59:01.028057 140088353853824 dataset_builder.py:80] Number of filenames to read: 1 WARNING:tensorflow:num_readers has been reduced to 1 to match input file shards. W0912 15:59:01.028110 140088353853824 dataset_builder.py:86] num_readers has been reduced to 1 to match input file shards. WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/object_detection/builders/dataset_builder.py:100: parallel_interleave (from tensorflow.python.data.experimental.ops.interleave_ops) is deprecated and will be removed in a future version. Instructions for updating: Use tf.data.Dataset.interleave(map_func, cycle_length, block_length, num_parallel_calls=tf.data.AUTOTUNE) instead. If sloppy execution is desired, use tf.data.Options.deterministic. W0912 15:59:01.029739 140088353853824 deprecation.py:350] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/object_detection/builders/dataset_builder.py:100: parallel_interleave (from tensorflow.python.data.experimental.ops.interleave_ops) is deprecated and will be removed in a future version. Instructions for updating: Use tf.data.Dataset.interleave(map_func, cycle_length, block_length, num_parallel_calls=tf.data.AUTOTUNE) instead. If sloppy execution is desired, use tf.data.Options.deterministic. WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/object_detection/builders/dataset_builder.py:235: DatasetV1.map_with_legacy_function (from tensorflow.python.data.ops.dataset_ops) is deprecated and will be removed in a future version. Instructions for updating: Use tf.data.Dataset.map() W0912 15:59:01.047504 140088353853824 deprecation.py:350] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/object_detection/builders/dataset_builder.py:235: DatasetV1.map_with_legacy_function (from tensorflow.python.data.ops.dataset_ops) is deprecated and will be removed in a future version. Instructions for updating: Use tf.data.Dataset.map() WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: sparse_to_dense (from tensorflow.python.ops.sparse_ops) is deprecated and will be removed in a future version. Instructions for updating: Create a tf.sparse.SparseTensor and use tf.sparse.to_dense instead. W0912 15:59:04.354701 140088353853824 deprecation.py:350] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: sparse_to_dense (from tensorflow.python.ops.sparse_ops) is deprecated and will be removed in a future version. Instructions for updating: Create a tf.sparse.SparseTensor and use tf.sparse.to_dense instead. WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: to_float (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version. Instructions for updating: Use tf.cast instead. W0912 15:59:05.460015 140088353853824 deprecation.py:350] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: to_float (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version. Instructions for updating: Use tf.cast instead. INFO:tensorflow:Waiting for new checkpoint at ./training_outlook_action_ctx/training_1 I0912 15:59:07.470988 140088353853824 checkpoint_utils.py:136] Waiting for new checkpoint at ./training_outlook_action_ctx/training_1 INFO:tensorflow:Found new checkpoint at ./training_outlook_action_ctx/training_1/ckpt-2 I0912 15:59:07.471778 140088353853824 checkpoint_utils.py:145] Found new checkpoint at ./training_outlook_action_ctx/training_1/ckpt-2 /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/keras/backend.py:450: UserWarning: tf.keras.backend.set_learning_phase is deprecated and will be removed after 2020-10-11. To update it, simply pass a True/False value to the training argument of the __call__ method of your layer or model. warnings.warn('tf.keras.backend.set_learning_phase is deprecated and ' INFO:tensorflow:depth of additional conv before box predictor: 0 I0912 15:59:14.530708 140088353853824 convolutional_keras_box_predictor.py:152] depth of additional conv before box predictor: 0 WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: calling crop_and_resize_v1 (from tensorflow.python.ops.image_ops_impl) with box_ind is deprecated and will be removed in a future version. Instructions for updating: box_ind is deprecated, use box_indices instead W0912 15:59:19.416570 140088353853824 deprecation.py:554] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: calling crop_and_resize_v1 (from tensorflow.python.ops.image_ops_impl) with box_ind is deprecated and will be removed in a future version. Instructions for updating: box_ind is deprecated, use box_indices instead WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/autograph/impl/api.py:459: Tensor.experimental_ref (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version. Instructions for updating: Use ref() instead. W0912 15:59:19.930708 140088353853824 deprecation.py:350] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/autograph/impl/api.py:459: Tensor.experimental_ref (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version. Instructions for updating: Use ref() instead. WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: softmax_cross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and will be removed in a future version. Instructions for updating:

Future major versions of TensorFlow will allow gradients to flow into the labels input on backprop by default.

See tf.nn.softmax_cross_entropy_with_logits_v2.

W0912 15:59:23.301795 140088353853824 deprecation.py:350] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: softmax_cross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and will be removed in a future version. Instructions for updating:

Future major versions of TensorFlow will allow gradients to flow into the labels input on backprop by default.

See tf.nn.softmax_cross_entropy_with_logits_v2.

WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: batch_gather (from tensorflow.python.ops.array_ops) is deprecated and will be removed after 2017-10-25. Instructions for updating: tf.batch_gather is deprecated, please use tf.gather with batch_dims=tf.rank(indices) - 1 instead. W0912 15:59:28.054638 140088353853824 deprecation.py:350] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: batch_gather (from tensorflow.python.ops.array_ops) is deprecated and will be removed after 2017-10-25. Instructions for updating: tf.batch_gather is deprecated, please use tf.gather with batch_dims=tf.rank(indices) - 1 instead. 2022-09-12 15:59:36.943932: I tensorflow/stream_executor/cuda/cuda_dnn.cc:384] Loaded cuDNN version 8401 WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: to_int64 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version. Instructions for updating: Use tf.cast instead. W0912 15:59:39.123106 140088353853824 deprecation.py:350] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1082: to_int64 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version. Instructions for updating: Use tf.cast instead. INFO:tensorflow:Finished eval step 0 I0912 15:59:39.147901 140088353853824 model_lib_v2.py:966] Finished eval step 0 WARNING:tensorflow:From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/autograph/impl/api.py:459: py_func (from tensorflow.python.ops.script_ops) is deprecated and will be removed in a future version. Instructions for updating: tf.py_func is deprecated in TF V2. Instead, there are two options available in V2. - tf.py_function takes a python function which manipulates tf eager tensors instead of numpy arrays. It's easy to convert a tf eager tensor to an ndarray (just call tensor.numpy()) but having access to eager tensors means tf.py_functions can use accelerators such as GPUs as well as being differentiable using a gradient tape. - tf.numpy_function maintains the semantics of the deprecated tf.py_func (it is not differentiable, and manipulates numpy arrays). It drops the stateful argument making all functions stateful.

W0912 15:59:39.260100 140088353853824 deprecation.py:350] From /home/robotiq-c3po/anaconda3/envs/tf_2/lib/python3.9/site-packages/tensorflow/python/autograph/impl/api.py:459: py_func (from tensorflow.python.ops.script_ops) is deprecated and will be removed in a future version. Instructions for updating: tf.py_func is deprecated in TF V2. Instead, there are two options available in V2. - tf.py_function takes a python function which manipulates tf eager tensors instead of numpy arrays. It's easy to convert a tf eager tensor to an ndarray (just call tensor.numpy()) but having access to eager tensors means tf.py_functions can use accelerators such as GPUs as well as being differentiable using a gradient tape. - tf.numpy_function maintains the semantics of the deprecated tf.py_func (it is not differentiable, and manipulates numpy arrays). It drops the stateful argument making all functions stateful.

INFO:tensorflow:Performing evaluation on 168 images. I0912 16:00:17.393700 140088353853824 coco_evaluation.py:293] Performing evaluation on 168 images. creating index... index created! INFO:tensorflow:Loading and preparing annotation results... I0912 16:00:17.397668 140088353853824 coco_tools.py:116] Loading and preparing annotation results... INFO:tensorflow:DONE (t=0.02s) I0912 16:00:17.413546 140088353853824 coco_tools.py:138] DONE (t=0.02s) creating index... index created! Running per image evaluation... Evaluate annotation type bbox DONE (t=0.71s). Accumulating evaluation results... DONE (t=0.14s). Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000 INFO:tensorflow:Eval metrics at step 1000 I0912 16:00:18.290414 140088353853824 model_lib_v2.py:1015] Eval metrics at step 1000 INFO:tensorflow: + DetectionBoxes_Precision/mAP: 0.000000 I0912 16:00:18.291919 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Precision/mAP: 0.000000 INFO:tensorflow: + DetectionBoxes_Precision/[email protected]: 0.000000 I0912 16:00:18.292804 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Precision/[email protected]: 0.000000 INFO:tensorflow: + DetectionBoxes_Precision/[email protected]: 0.000000 I0912 16:00:18.293548 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Precision/[email protected]: 0.000000 INFO:tensorflow: + DetectionBoxes_Precision/mAP (small): 0.000000 I0912 16:00:18.294231 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Precision/mAP (small): 0.000000 INFO:tensorflow: + DetectionBoxes_Precision/mAP (medium): 0.000000 I0912 16:00:18.294910 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Precision/mAP (medium): 0.000000 INFO:tensorflow: + DetectionBoxes_Precision/mAP (large): 0.000000 I0912 16:00:18.295593 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Precision/mAP (large): 0.000000 INFO:tensorflow: + DetectionBoxes_Recall/AR@1: 0.000000 I0912 16:00:18.296273 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Recall/AR@1: 0.000000 INFO:tensorflow: + DetectionBoxes_Recall/AR@10: 0.000000 I0912 16:00:18.296947 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Recall/AR@10: 0.000000 INFO:tensorflow: + DetectionBoxes_Recall/AR@100: 0.000000 I0912 16:00:18.297632 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Recall/AR@100: 0.000000 INFO:tensorflow: + DetectionBoxes_Recall/AR@100 (small): 0.000000 I0912 16:00:18.298316 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Recall/AR@100 (small): 0.000000 INFO:tensorflow: + DetectionBoxes_Recall/AR@100 (medium): 0.000000 I0912 16:00:18.299822 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Recall/AR@100 (medium): 0.000000 INFO:tensorflow: + DetectionBoxes_Recall/AR@100 (large): 0.000000 I0912 16:00:18.301264 140088353853824 model_lib_v2.py:1018] + DetectionBoxes_Recall/AR@100 (large): 0.000000 INFO:tensorflow: + Loss/RPNLoss/localization_loss: 0.862616 I0912 16:00:18.301923 140088353853824 model_lib_v2.py:1018] + Loss/RPNLoss/localization_loss: 0.862616 INFO:tensorflow: + Loss/RPNLoss/objectness_loss: 0.017968 I0912 16:00:18.302563 140088353853824 model_lib_v2.py:1018] + Loss/RPNLoss/objectness_loss: 0.017968 INFO:tensorflow: + Loss/BoxClassifierLoss/localization_loss: 0.000000 I0912 16:00:18.303211 140088353853824 model_lib_v2.py:1018] + Loss/BoxClassifierLoss/localization_loss: 0.000000 INFO:tensorflow: + Loss/BoxClassifierLoss/classification_loss: 0.002990 I0912 16:00:18.303851 140088353853824 model_lib_v2.py:1018] + Loss/BoxClassifierLoss/classification_loss: 0.002990 INFO:tensorflow: + Loss/regularization_loss: 0.000000 I0912 16:00:18.304493 140088353853824 model_lib_v2.py:1018] + Loss/regularization_loss: 0.000000 INFO:tensorflow: + Loss/total_loss: 0.883575 I0912 16:00:18.305129 140088353853824 model_lib_v2.py:1018] + Loss/total_loss: 0.883575 INFO:tensorflow:Waiting for new checkpoint at ./training_outlook_action_ctx/training_1

3. Steps to reproduce

I start training with the command: python model_main_tf2.py --pipeline_config_path=./training_outlook_action_ctx/training_1/pipeline.config --model_dir=./training_outlook_action_ctx/training_1 --alsologtostderr After closing training i start evaluation with: python model_main_tf2.py --pipeline_config_path=./training_outlook_action_ctx/training_1/pipeline.config --model_dir=./training_outlook_action_ctx/training_1 --checkpoint_dir=./training_outlook_action_ctx/training_1 --alsologtostderr

Values during training: I0912 15:37:09.102276 140574688362880 model_lib_v2.py:708] {'Loss/BoxClassifierLoss/classification_loss': 0.114061415, 'Loss/BoxClassifierLoss/localization_loss': 0.09612781, 'Loss/RPNLoss/localization_loss': 0.16708645, 'Loss/RPNLoss/objectness_loss': 0.0012185145, 'Loss/regularization_loss': 0.0, 'Loss/total_loss': 0.3784942, 'learning_rate': 0.0026}

4. Expected behavior

I expect at least some values, even if they are very, very small.

5. Additional context

Here is my pipeline.config file:

`Faster R-CNN with Resnet Sync-trained on COCO (8 GPUs), Initialized from Imagenet classification checkpoint TF2-Compatible, Not TPU-Compatible

model { faster_rcnn { num_classes: 7 image_resizer { keep_aspect_ratio_resizer { min_dimension: 800 max_dimension: 1333 pad_to_max_dimension: true } } feature_extractor { type: 'faster_rcnn_resnet101_keras' } first_stage_anchor_generator { grid_anchor_generator { scales: [0.25, 0.5, 1.0, 2.0] aspect_ratios: [0.5, 1.0, 2.0] height_stride: 16 width_stride: 16 } } first_stage_box_predictor_conv_hyperparams { op: CONV regularizer { l2_regularizer { weight: 0.0 } } initializer { truncated_normal_initializer { stddev: 0.01 } } } first_stage_nms_score_threshold: 0.0 first_stage_nms_iou_threshold: 0.7 first_stage_max_proposals: 300 first_stage_localization_loss_weight: 2.0 first_stage_objectness_loss_weight: 1.0 initial_crop_size: 14 maxpool_kernel_size: 2 maxpool_stride: 2 second_stage_box_predictor { mask_rcnn_box_predictor { use_dropout: false dropout_keep_probability: 1.0 fc_hyperparams { op: FC regularizer { l2_regularizer { weight: 0.0 } } initializer { variance_scaling_initializer { factor: 1.0 uniform: true mode: FAN_AVG } } } } } second_stage_post_processing { batch_non_max_suppression { score_threshold: 0.0 iou_threshold: 0.6 max_detections_per_class: 100 max_total_detections: 100 } score_converter: SOFTMAX } second_stage_localization_loss_weight: 2.0 second_stage_classification_loss_weight: 1.0 } } train_config: { batch_size: 2 num_steps: 200000 optimizer { momentum_optimizer: { learning_rate: { cosine_decay_learning_rate { learning_rate_base: 0.01 total_steps: 200000 warmup_learning_rate: 0.0 warmup_steps: 5000 } } momentum_optimizer_value: 0.9 } use_moving_average: false } gradient_clipping_by_norm: 10.0 fine_tune_checkpoint_version: V2 fine_tune_checkpoint: "/media/robotiq-c3po/HARD1T/Tensorflow2/models/research/object_detection/pretrained_models/faster_rcnn_resnet101_v1_800x1333_coco17_gpu-8/checkpoint/ckpt-0" fine_tune_checkpoint_type: "detection" data_augmentation_options { random_horizontal_flip { } } data_augmentation_options { random_adjust_hue { } } data_augmentation_options { random_adjust_contrast { } } data_augmentation_options { random_adjust_saturation { } } data_augmentation_options { random_square_crop_by_scale { scale_min: 0.6 scale_max: 1.3 } } } train_input_reader: { label_map_path: "./training_outlook_action_ctx/data/label_map.pbtxt" tf_record_input_reader { input_path: "./training_outlook_action_ctx/data/train.records" } } eval_config: { metrics_set: "coco_detection_metrics" use_moving_averages: false batch_size: 2; } eval_input_reader: { label_map_path: "./training_outlook_action_ctx/data/label_map.pbtxt" shuffle: false num_epochs: 2 tf_record_input_reader { input_path: "./training_outlook_action_ctx/data/val.records" } }`

6. System information

• OS Platform and Distribution: Debian GNU/Linux 11 (bullseye)
• TensorFlow installed from (source or binary): https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/install.html
• TensorFlow version (use command below):v2.9.0-18-gd8ce9f9c301 2.9.1
• Python version: 3.9.12
• CUDA/cuDNN version:CUDA Version: 11.7,
• GPU model and memory:2x NVIDIA Corporation GP104 [GeForce GTX 1070]

Thanks in advance

[samchapman94]

I am suffering from similar issues

[berniecamus]

+1 from me on FasterRCNN. I have a similar setup and data using SSD and it works, but with FasterRCNN I'm getting all Zeros.

[Petros626]

Please don't post all these information, it doesn't help us. Post only the specific parts and necessary to guess where the issue is.

[TahaErr]

Hi, do you find any solution? @samchapman94 @Edi2410

[tamaratoma]

I am also having the same issue. When I train my model only with Yolov8, it does show me the correct mAP, but when I use Keras, I always get 0

[TahaErr]

@tamaratoma @Edi2410 @samchapman94 Hi, I find the soultion. I just increased the number of training steps. it was like 2k and ı changed to 10k. Now, I'm getting 0.5 mAP50 values kinda bad but better than 0.

[google-ml-butler[bot]]

Are you satisfied with the resolution of your issue? Yes No