metrics_df = pd.read_json(
        "/kaggle/working/metrics (6).json", orient="records", lines=True)
mdf = metrics_df.sort_values("iteration")
print(mdf.tail(10).T)

# Plot loss
fig, ax = plt.subplots()

mdf1 = mdf[~mdf["total_loss"].isna()]
ax.plot(mdf1["iteration"], mdf1["total_loss"], c="C0", label="train")

if "validation_loss" in mdf.columns:
    mdf2 = mdf[~mdf["validation_loss"].isna()]
    ax.plot(mdf2["iteration"], mdf2["validation_loss"],
            c="C1", label="validation")

ax.legend()
ax.set_title("Loss curve")
plt.show()

# Plot Accuracy stage 0 fastrcnn
fig, ax = plt.subplots()

mdf1 = mdf[~mdf["mask_rcnn/accuracy"].isna()]
ax.plot(mdf1["iteration"], mdf1["mask_rcnn/accuracy"],
        c="C0", label="train")

ax.legend()
ax.set_title("MASKRCNN Accuracy curve")

plt.show()

# Plot Accuracy maskrcnn

fig, ax = plt.subplots()

mdf1 = mdf[~mdf["stage0/fast_rcnn/cls_accuracy"].isna()]
ax.plot(mdf1["iteration"], mdf1["stage0/fast_rcnn/cls_accuracy"],
        c="C0", label="train")

ax.legend()
ax.set_title("FASTRCNN CLS Accuracy curve")
plt.show()



# Plot Bounding Box regressor loss
fig, ax = plt.subplots()

mdf1 = mdf[~mdf["loss_box_reg_stage0"].isna()]
ax.plot(mdf1["iteration"], mdf1["loss_box_reg_stage0"], c="C0", label="train")

ax.legend()
ax.set_title("loss_box_reg")
plt.show()




# Plot loss cls stage0
fig, ax = plt.subplots()

mdf1 = mdf[~mdf["loss_cls_stage0"].isna()]
ax.plot(mdf1["iteration"], mdf1["loss_cls_stage0"], c="C0", label="train")

ax.legend()
ax.set_title("loss_cls_stage0 ")
plt.show()


# Plot loss mask
fig, ax = plt.subplots()

mdf1 = mdf[~mdf["loss_mask"].isna()]
ax.plot(mdf1["iteration"], mdf1["loss_mask"], c="C0", label="train")

ax.legend()
ax.set_title("loss mask")
plt.show()