Annie

> The learning rate set in the [TF2 SSD MobileNet V2 config file](https://github.com/tensorflow/models/blob/master/research/object_detection/configs/tf2/ssd_mobilenet_v2_320x320_coco17_tpu-8.config) is 10x higher than that of the other SSD MobileNet models. This causes loss during training to...

> Hi @Annieliaquat , yes I did! They tried to implement a change to fix it, but it got rejected. #10531 > > The high learning rate is intended for...

Is this issue solved now?? CenterNet MobileNetV2 FPN 512x512 is trainable now?

Why is not anyone answering the issues?

> image = decoded_dict[fields.InputDataFields.image] > img = Image.open(io.BytesIO(image)).convert("RGB") > input_tensor = tf.convert_to_tensor(img) > input_tensor = input_tensor[tf.newaxis, ...] I have done the way you have said. But I still face this...

> Thank you very much i created a fork with your fix and its now working ♥ there was some numpy error but i added a line and it got...

> > > Thank you very much i created a fork with your fix and its now working ♥ there was some numpy error but i added a line and...

> it should be > > ``` > if decoded_dict: > image = decoded_dict[fields.InputDataFields.image] > img = Image.open(io.BytesIO(image)).convert("RGB") > img = np.asarray(img) > input_tensor = tf.convert_to_tensor(img) > input_tensor = input_tensor[tf.newaxis,...

> Thank you very much i created a fork with your fix and its now working ♥ there was some numpy error but i added a line and it got...

> it should be > > ``` > if decoded_dict: > image = decoded_dict[fields.InputDataFields.image] > img = Image.open(io.BytesIO(image)).convert("RGB") > img = np.asarray(img) > input_tensor = tf.convert_to_tensor(img) > input_tensor = input_tensor[tf.newaxis,...